Imaging

Ophthalmol Clin N Am 17 (2004) ix–x

Preface

Imaging

Joel S. Schuman, MD

Guest Editor

This issue of the Ophthalmology Clinics of North tumor evaluation, especially tumors of the ciliary

America focuses on ophthalmic diagnostics, primarily

imaging, but including advanced functional measures

as well. The articles contained herein represent the

state-of-the-art in ophthalmic diagnostics and are

written by the leaders in the field. This issue has

been endorsed by the International Society for Imag-

ing in the Eye, an organization devoted to the

advancement of the science of ophthalmic imaging.

In ‘‘Optical Coherence Tomography of the Ante-

rior Segment of the Eye,’’ David Huang, MD, PhD,

Yan Li, MS, and Sunita Radhakrishnan, MD, explore

the usefulness of OCT in anterior segment evalua-

tion. The technology reveals in great detail the

structure of the cornea and anterior segment struc-

tures. This technology offers the advantage of non-

contact, noninvasive assessment of the anterior eye,

including anterior chamber angle evaluation and high

resolution measurement of corneal thickness. Refrac-

tive surgery planning and results can be investigated

as well, including LASIK flap and bed thickness

and morphology.

Somewhat more conventional high resolution an-

terior segment imaging is discussed in, ‘‘Anterior

Segment Imaging: Ultrasound Biomicroscopy,’’ by

Hiroshi Ishikawa, MD, and me. We review the

myriad clinical applications of ultrasound biomi-

croscopy, a technology that is currently available to

clinicians, unlike anterior segment OCT. Ultrasound

biomicroscopy has particular use in anterior segment

0896-1549/04/$ – see front matter D 2004 Elsevier Inc. All right

doi:10.1016/j.ohc.2004.02.001

body, and in angle closure glaucoma, where a rapid

ultrasound biomicroscopy dark room test can mea-

sure whether or not an angle is actually occludable.

This is especially helpful in patient education, and

this application should not be underestimated.

The wide spectrum of macular diseases is ex-

plored in ‘‘Clinical Applications of Optical Coher-

ence Tomography for the Diagnosis and Management

of Macular Diseases’’ by Irene Voo, MD, Elias C.

Mavrofrides, MD, and Carmen A. Puliafito, MD,

MBA. The cross-sectional visualization of the retina

made possible by OCT is used to full advantage in

this article. OCT enables near-histologic evaluation of

the macula, revealing the nature and location of

pathophysiology for the clinician.

In ‘‘Imaging in Glaucoma,’’ Daniel M. Stein,

BA, Gadi Wollstein, MD, and I review the various

imaging technologies used to evaluate glaucoma. The

devices themselves are described in addition to their

clinical applications. Comparisons are given so that

the clinician can visualize the unique information

provided by each device. The clinical studies are

summarized comparing these devices as well.

Moving to functional measures, ‘‘Electrophysio-

logic Imaging of Retinal and Optic Nerve Damage:

The Multifocal Technique’’ by Donald Hood, PhD,

is an in-depth review of multifocal electrophysio-

logic evaluation. The promise of these new methods

is tremendous, with the potential of objective, quanti-

s reserved.

J.S. Schuman / Ophthalmol Clin N Am 17 (2004) ix–xx

tative measurement of functional abnormalities. This

is in stark contrast to perimetry, which involves the

subjective interpretation of a subjective test.

Continuing in the objective, quantitative electro-

physiologic vein, but expanding to include structural

assessment as well, Thomas R. Hedges III, MD, and

Maria-Luz Amaro Quereza, OD, investigate the use

of state-of-the-art technologies in neuro-ophthalmic

disease in ‘‘Multifocal Visual Evoked Potential, Mul-

tifocal Electroretinography, and Optical Coherence

Tomography in the Diagnosis of Subclinical Loss of

Vision.’’ This article discusses a variety of patholo-

gies and portrays the clinical value of these currently

available methods.

Perhaps the most cutting-edge article in this

issue, ‘‘High-Resolution Functional Optical Imaging:

From the Neocortex to the Eye,’’ by Amiram Grin-

vald, PhD, Tobias Bonhoeffer, PhD, Ivo Vanzetta,

PhD, Ayala Pollack, MD, Eyal Aloni, MD, Ron Ofri,

MD, and Darin Nelson, PhD, reviews emerging tech-

nology for measurement of ocular blood flow and

oximetry. The potential of the technology described

is enormous, and this section promises to translate

what was once science fiction to science.

Joel S. Schuman, MD

University of Pittsburgh School of Medicine

UPMC Eye Center

203 Lothrop Street

Eye and Ear Institute, Suite 816

Pittsburgh, PA 15213, USA

E-mail address: [email protected]

Ophthalmol Clin N Am 17 (2004) 1–6

Optical coherence tomography of the anterior segment of

the eye

David Huang, MD, PhD*, Yan Li, MS, Sunita Radhakrishnan, MD

Cole Eye Institute, The Cleveland Clinic, Department of Neurosciences NC3, 9500 Euclid Avenue, I20,

Cleveland, OH 44195, USA

Optical coherence tomography (OCT) [1] is a tical line to the OCT image. A corneal image pro-

novel cross-sectional and three-dimensional imaging

modality that uses low-coherence interferometry to

achieve axial (depth) resolutions in the range of 3 to

20 mm. OCT has several theoretical advantages

when compared with current imaging modalities for

imaging the anterior segment of the eye. Unlike

ultrasound, OCT employs light; therefore, it does

not require fluid immersion or probe contact. Further-

more, OCT has a spatial resolution that easily sur-

passes that of even ultrahigh-frequency ultrasound

[2–9]. Although confocal scanning microscopy

[10–15] can obtain even higher resolution than

OCT, it requires short focal distances and can image

only a small area of the eye at a time. OCT uses

interferometry for depth resolution; therefore, it can

have a long working distance and a wide field of

transverse scanning. Slit illumination imaging, such

as Scheimpflug photography [16] or the Orbscan

system [17–21] (Bausch & Lomb, Rochester, New

York), can also provide cross-sectional images, but

accurate biometry is difficult. The angles between the

slit illumination, subject anatomic surfaces, and im-

aging axis require complex computations for anatom-

ic reconstruction and measurements. OCT detects

retroreflected light and is more amenable to simple

interpretation and accurate measurements.

The primary limitations of OCT imaging of the

anterior segment are speed and penetration. Commer-

cial retinal scanners (Carl Zeiss Meditec, Dublin,

California) have a scanning speed of 100 to 400 axial

scans per second. Each axial scan contributes a ver-


doi:10.1016/S0896-1549(03)00103-2

* Corresponding author.

E-mail address: [email protected] (D. Huang).

duced by the Zeiss OCT 1 system (Fig. 1) has visible

motion artifacts because the image is acquired over

1 second. It also has coarse pixel grain because it

contains only 100 vertical lines. The 0.8-mm wave-

length employed for retinal scanning is close to

visible wavelengths and cannot directly visualize

angle structures owing to scattering loss through the

limbus [22]. A widely useful anterior segment OCT

system would need to overcome both limitations.

This article presents the results using a high-speed

OCT prototype developed collaboratively between

Professor Joseph Izatt (Duke University), Professor

Andrew Rollins (Case University), and the authors at

the Cleveland Clinic Foundation. The system has a

scan rate of 4000 lines per second, which is fast

enough for biometric applications. It uses a longer

wavelength of 1.3 mm, which decreases scattering

through turbid tissues and allows visualization of

angle structures [23]. The system is useful in a range

of clinical applications from laser-assisted in situ

keratomileusis (LASIK) surgery to narrow angle

glaucoma. OCT of the anterior segment is still in its

infancy; therefore, it is discussed at the end of this

article in a section on future prospects.

Background

The precursor technology of OCT [1], optical

coherence domain reflectometry (OCDR), had its

earliest demonstration in biomedical applications

with the measurement of corneal thickness [24].

OCDR is an optical ranging technique, or a method

for measuring the distance between the measurement

s reserved.

Fig. 1. OCT of the cornea with 6.5-mm scan width

performed with the Zeiss OCT 1 system. The image has

not been corrected for aspect ratio and the divergent beam

scan path; therefore, the thickness and corneal curvature

are exaggerated.

D. Huang et al / Ophthalmol Clin N Am 17 (2004) 1–62

system and a target. In OCDR, an optical probe beam

is directed toward a target sample, and the reflected

light is recombined with a reference reflection in an

interferometer. Interference occurs when the two

reflections are mutually coherent. Because OCDR

uses light with a short coherence length, coherence

occurs only when the reference and sample reflections

are closely matched in propagation delay. By scan-

ning the delay of the reference reflection and record-

ing the resulting fluctuation in interference signal, the

OCDR system resolves the amplitudes of sample

reflections as a function of depth (axial delay). OCT

is based on OCDR, with the addition of transverse

scanning of the probe beam to provide at least an-

other spatial dimension for imaging.

Izatt et al published the first report of OCT for

corneal and anterior segment imaging in 1994 [22].

After that report, little attention was paid to ante-

rior segment applications until the Lubeck group

described OCT imaging of laser thermokeratoplasty

lesions in 1997 [25,26], and Maldonado et al [27–29]

reported imaging of the LASIK flap in 1998. Since

then, the rapid popularization of corneal refractive

surgery has spurred investigators to apply OCT to

corneal imaging and to refine the instrumentation

for anterior segment OCT.

In most published reports of anterior segment

OCT applications, the commercial retinal OCT scan-

ner has been used [27–44]. Although visualization of

corneal surgical anatomy and thickness measurements

are possible, the imaging speed, penetration, and field

of view are very limited with the retinal scanner.

Ocular imaging with the 1.3-mm wavelength was

first reported by Radhakrishnan et al using a system

developed by Izatt’s group [23]. The Izatt system

takes full advantage of the higher power that can be

used safely at the longer wavelength to achieve a

high acquisition rate of 4000 axial scans per second.

This high-speed OCT engine has been used in several

anterior segment OCT scanners to generate the clini-

cal results discussed herein.

High-speed corneal and anterior segment optical

coherence tomography at 1.3-Mm wavelength

Using a longer wavelength of 1.3 mm for corneal

and anterior segment (CAS) OCT provides important

advantages when compared with the 0.8-mm wave-

length commonly used for retinal imaging. Because

scattering loss is much lower at the longer wave-

length, 1.3-mm OCT can penetrate the limbus and

sclera to provide a view of the angle. The absorption

by water is also much stronger at 1.3 mm (9.3-mm

absorption length) [45]. For an eye of average length,

91% of 1.3-mm light that falls on the cornea will be

absorbed by the ocular media, leaving only 9% to

reach the retina. This absorption allows the use of

much higher optical power without damaging the

retina. The permissible exposure level at the 1.3-mmwavelength is 15 mW according to the current stan-

dard set by the American Laser Institute and the

American National Standards Institute (ANSI 2000)

[46]. This level is 20 times higher than the 0.7-mW

limit at the 0.8-mm wavelength [46]. As a result,

CAS OCT can use much higher power and achieve

much higher scan rates. The authors developed sev-

eral CAS OCT systems using a high-speed OCT en-

gine (4000 A-scans per second) with a 1.3-mm super

luminescent diode (SLD) light source.

The final version of the CAS OCT scanner was

mounted on a slit lamp base and used a charge couple

device (CCD) camera to visualize the scan area in real

time. The scan geometry was telecentric (rectangular),

with adjustable scan widths of 4 to 15 mm and scan

depths of 3.25 or 6 mm. Eight image frames were

acquired and displayed per second in real time, each

with 500 axial scans. The axial resolution was 14 mmfull-width-half-maximum in cornea. The main advan-

tage of the telecentric system was its wide field

capability, which was essential for corneal and ante-

rior chamber studies. This system was used for all of

the studies described herein except for the initial angle

assessment study.

Laser-assisted in situ keratomileusis anatomy

The 1.3-mm wide-field CAS OCT system was

used to examine post-LASIK eyes (Fig. 2). The scan

dimensions are 12 mm wide and 3.25 mm deep

Fig. 2. OCT of an eye 1 day after LASIK surgery.

Fig. 3. High-speed 1.3-mmwavelength OCTof an open angle.

D. Huang et al / Ophthalmol Clin N Am 17 (2004) 1–6 3

(in air). When compared with the corneal image

(see Fig. 1) obtained with a 0.8-mm wavelength retinal

scanner, there are several improvements. There is no

apparent motion artifact at the higher image acquisi-

tion rate of eight frames per second. The image detail

is much finer with 500 axial scans per frame com-

pared with 100 axial scans per frame. The telecentric

design allows the capture of corneal details over a

much wider scan width.

Many useful anatomic features relevant to LASIK

surgery can be identified in the image (Fig. 2). The

anterior surface reflection is strong at the perpen-

dicular incidence and produces a vertical flare that

defines the corneal apex. Finer features, such as the

epithelial–Bowman and the flap lamellar boundaries,

are best visualized with a slight off-normal beam

incidence angle in the midperiphery. The flap internal

reflectivity is stronger than that of the posterior

stroma. The thickness of the cornea, flap, and poste-

rior stromal bed can be measured accurately from

the tomograph.

Angle assessment

Gonioscopy is the gold standard for evaluating the

anterior chamber angle; however, it is highly subjec-

tive and requires specialized training. Cross-sectional

imaging of the anterior chamber angle with ultra-

sound biomicroscopy or OCT is easier to interpret.

Furthermore, objective quantification of the angle can

be obtained from a cross section. Ultrasound biomi-

croscopy [47] and Scheimpflug photography [48]

have been used for quantitative angle evaluation.

OCT can provide the same detailed angle anatomy

with the added advantage of being noncontact and

easy to perform.

A 1.3-mm high-speed CAS OCT system was used

to assess the angle width. Computer image processing

was performed to obtain correctly dimensioned

images, with adjustments for the scan geometry and

refraction of the OCT beam at the anterior eye sur-

face. Images of an open angle and an occludable an-

gle are shown in Figs. 3 and 4, respectively. Corneal,

scleral, and iris anatomy are visualized in detail.

Features in the limbus and angle are clearly shown,

including the scleral spur, ciliary body band, angle

recess, and iris root. For angle measurements, it is

particularly useful that the scleral spur is highly

reflective and easily identified on OCT.

A clinical study was performed to compare OCT

and ultrasound biomicroscopy angle parameters with

gonioscopic grading by glaucoma specialists. A total

of 31 eyes in 28 subjects were examined. Eight eyes

were judged to be occludable on gonioscopy. Sub-

jects underwent OCT and ultrasound biomicroscopy

imaging of the nasal and temporal anterior chamber

angles. OCT and ultrasound biomicroscopy had ex-

cellent correlation with gonioscopy in terms of the

identification of occludable angles. The best OCT

parameters were slightly better than ultrasound bio-

microscopy parameters, with 100% sensitivity and

95.7% specificity for detecting gonioscopically oc-

cludable angles.

Anterior chamber width and other biometric

parameters

Optical coherence tomography is well suited for

ocular biometry owing to its high image resolution.

When compared with ultrasound, the noncontact

nature of OCT eliminates discomfort and distortion

from probe contact or immersion. The authors’ high-

speed wide-field OCT prototype was able to produce

detailed images of the entire anterior chamber without

any visible motion artifact (Fig. 5). This characteristic

Fig. 4. High-speed 1.3-mm wavelength OCT of an occlud-

able angle.


makes the accurate measurement of anterior cham-

ber width and other biometric parameters possible.

Anterior chamber width measurement is clini-

cally important for sizing angle-supported anterior

chamber intraocular lenses. With the increasing use

of refractive phakic intraocular lenses, accurate sizing

becomes an important issue. An intraocular lens that

is too large can press on the iris root and produce pupil

ovalization; an intraocular lens that is too small can

lead to lens movement, decentration, corneal endo-

thelial damage, and iritis [49]. The traditional method

for sizing uses the external corneal diameter, which

is assumed to correspond to the internal anterior

chamber width. Typically, the intraocular lens length

is chosen to be the corneal diameter plus a constant

such as 1 mm.

When the wide-field CAS OCT system was used,

it was possible to measure the internal width of the

anterior chamber directly (Fig. 5). Anterior chamber

width as measured by OCT was compared with

corneal diameter measured by a Holladay cornea

gauge in 20 normal subjects. The anterior chamber

width was 12.53 F 0.47 mm (mean F SD). The

difference of the anterior chamber width from the

corneal diameter was 0.75 F 0.44 mm with a range

Fig. 5. Anterior chamber (AC) imaged with the wide-field (15-mm

system. The AC width is measured between angle recesses; its de

of 1.84 mm. The improvement of the SD from 0.47

to 0.44 mm was minimal and showed that intraocular

lens sizing using the corneal diameter would be only

marginally better than using the same size for all eyes.

OCT can improve the accuracy of intraocular lens

sizing several fold. The reproducibility of anterior

chamber width measurements from OCT images was

assessed by an analysis of variance. The variation of

anterior chamber width between images was small

(SD = 0.10 mm), but disagreement between the three

human graders was larger (SD = 0.29 mm). The

authors are developing an automated anterior cham-

ber width measurement software that will remove

the need for a human grader to place cursors at the

angle recesses. Anterior chamber depth and the crys-

talline lens vault were also measured from the OCT

images with high reproducibility.

Optical coherence tomography appears to be a

reproducible, convenient, and noncontact technique

to perform biometry of anterior chamber dimen-

sions. Further studies are needed to determine whether

it can contribute to a reduction of complications

through better fitting of intraocular lenses for the

anterior chamber.

Future developments

Optical coherence topography is a versatile tool

for visualization and measurement of corneal and an-

terior segment anatomy. It has the potential for im-

proving the functions currently served by Placido-ring

corneal topography, slit-scanning corneal topography,

ultrasound imaging, and ultrasound pachymetry.

Keratorefractive surgery, anterior chamber bio-

metry, and angle assessment are some of the appli-

cations that should benefit from CAS OCT. The

commercialization and general availability of this

technology should increase applications through the

ingenuity of many practitioners.

) setting on the slit lamp–mounted high-speed CAS OCT

pth is measured from the corneal apex to lens apex.

D. Huang et al / Ophthalmol Clin N Am 17 (2004) 1–6 5

References

[1] Huang D, Swanson EA, Lin CP, et al. Optical coher-

ence tomography. Science 1991;254(5035):1178–81.

[2] Allemann N, Chamon W, Silverman RH, et al. High-

frequency ultrasound quantitative analyses of corneal

scarring following excimer laser keratectomy. Arch

Ophthalmol 1993;111(7):968–73.

[3] Pavlin CJ, Easterbrook M, Harasiewicz K, Foster FS.

An ultrasound biomicroscopic analysis of angle-closure

glaucoma secondary to ciliochoroidal effusion in IgA

nephropathy. Am J Ophthalmol 1993;116(3):341–5.

[4] Reinstein DZ, Silverman RH, Coleman DJ. High-fre-

quency ultrasound measurement of the thickness of

the corneal epithelium. Refract Corneal Surg 1993;

9(5):385–7.

[5] Reinstein DZ, Silverman RH, Trokel SL, Allemann N,

Coleman DJ. High-frequency ultrasound digital signal

processing for biometry of the cornea in planning

phototherapeutic keratectomy [letter] [published erra-

tum appears in Arch Ophthalmol 1993 Jul;111(7):926].

Arch Ophthalmol 1993;111(4):430–1.

[6] Reinstein DZ, Silverman RH, Trokel SL, Coleman DJ.

Corneal pachymetric topography. Ophthalmology

1994;101(3):432–8.

[7] Reinstein DZ, Silverman RH, Rondeau MJ, Coleman

DJ. Epithelial and corneal thickness measurements by

high-frequency ultrasound digital signal processing.

Ophthalmology 1994;101(1):140–6.

[8] Riley SF, Nairn JP, Maestre FA, Smith TJ. Analysis

of the anterior chamber angle by gonioscopy and by

ultrasound biomicroscopy. Int Ophthalmol Clin 1994;

34(3):271–82.

[9] Reinstein DZ, Silverman RH, Sutton HF, Coleman

DJ. Very high-frequency ultrasound corneal analysis

identifies anatomic correlates of optical complica-

tions of lamellar refractive surgery: anatomic diagno-

sis in lamellar surgery. Ophthalmology 1999;106(3):

474–82.

[10] Jalbert I, Stapleton F, Papas E, Sweeney DF, Coroneo

M. In vivo confocal microscopy of the human cornea.

Br J Ophthalmol 2003;87(2):225–36.

[11] Bauer NJ, Wicksted JP, Jongsma FH, March WF, Hen-

drikse F, Motamedi M. Noninvasive assessment of

the hydration gradient across the cornea using confo-

cal Raman spectroscopy. Invest Ophthalmol Vis Sci

1998;39(5):831–5.

[12] Lemp MA, Dilly PN, Boyde A. Tandem-scanning

(confocal) microscopy of the full-thickness cornea.

Cornea 1985;4(4):205–9.

[13] Masters BR, Farmer MA. Three-dimensional confocal

microscopy and visualization of the in situ cornea.

Comput Med Imaging Graph 1993;17(3):211–9.

[14] Masters BR, Paddock S. In vitro confocal imaging of

the rabbit cornea. J Microsc 1990;158(Pt 2):267–74.

[15] Ichijima H, Petroll WM, Jester JV, Cavanagh HD.

Confocal microscopic studies of living rabbit cornea

treated with benzalkonium chloride. Cornea 1992;

11(3):221–5.

[16] Boker T, Sheqem J, Rauwolf M, Wegener A. Ante-

rior chamber angle biometry: a comparison of Scheim-

pflug photography and ultrasound biomicroscopy.

Ophthalmic Res 1995;27(Suppl 1):104–9.

[17] Yaylali V, Kaufman SC, Thompson HW. Corneal

thickness measurements with the Orbscan Topog-

raphy System and ultrasonic pachymetry. J Cataract

Refract Surg 1997;23(9):1345–50.

[18] Lattimore Jr MR, Kaupp S, Schallhorn S, Lewis RT.

Orbscan pachymetry: implications of a repeated mea-

sures and diurnal variation analysis. Ophthalmology

1999;106(5):977–81.

[19] Auffarth GU, Tetz MR, Biazid Y, Volcker HE. Measur-

ing anterior chamber depth with Orbscan Topography

System. J Cataract Refract Surg 1997;23(9):1351–5.

[20] Boscia F, La Tegola MG, Alessio G, Sborgia C. Accu-

racy of Orbscan optical pachymetry in corneas with

haze. J Cataract Refract Surg 2002;28(2):253–8.

[21] Cairns G, McGhee CN, Collins MJ, Owens H, Gamble

GD. Accuracy of orbscan II slit-scanning elevation to-

pography. J Cataract Refract Surg 2002;28(12):2181–7.

[22] Izatt JA, Hee MR, Swanson EA, et al. Micrometer-

scale resolution imaging of the anterior eye in vivo

with optical coherence tomography. Arch Ophthalmol

1994;112(12):1584–9.

[23] Radhakrishnan S, Rollins AM, Roth JE, et al. Real-time

optical coherence tomography of the anterior segment

at 1310 nm. Arch Ophthalmol 2001;119(8):1179–85.

[24] Huang D, Wang J, Lin CP, Puliafito CA, Fujimoto JG.

Micron-resolution ranging of cornea anterior chamber

by optical reflectometry. Lasers Surg Med 1991;11(5):

419–25.

[25] Koop N, Brinkmann R, Lankenau E, Flache S, Engel-

hardt R, Birngruber R. [Optical coherence tomography

of the cornea and the anterior eye segment]. Ophthal-

mologe 1997;94(7):481–6.

[26] Asiyo-Vogel MN, Koop N, Brinkmann R, et al. [Imag-

ing of laser thermokeratoplasty lesions by optical low

coherence tomography and polarization microscopy

after Sirius Red staining]. Ophthalmologe 1997;94(7):

487–91.

[27] Maldonado MJ. Undersurface ablation of the flap for

laser in situ keratomileusis retreatment. Ophthalmol-

ogy 2002;109(8):1453–64.

[28] Maldonado MJ, Munuera JM, Garcia-Layana A,

Moreno J, Aliseda D. Optical coherence tomography

(OCT) evaluation of the corneal cap and stromal bed

features after LASIK for high myopia. Presented at the

American Academy of Ophthalmology Annual Meet-

ing. New Orleans, November 8–11, 1998.

[29] Maldonado MJ, Ruiz-Oblitas L, Munuera JM, Aliseda

D, Garcia-Layana A, Moreno-Montanes J. Optical

coherence tomography evaluation of the corneal cap

and stromal bed features after laser in situ kerato-

mileusis for high myopia and astigmatism [In Pro-

cess Citation]. Ophthalmology 2000;107(1):81 – 7

[discussion: 88].

[30] Muscat S, McKay N, Parks S, Kemp E, Keating D.

Repeatability and reproducibility of corneal thickness


measurements by optical coherence tomography. In-

vest Ophthalmol Vis Sci 2002;43(6):1791–5.

[31] Neubauer AS, Priglinger SG, Thiel MJ, May CA,

Welge-Lussen UC. Sterile structural imaging of donor

cornea by optical coherence tomography. Cornea 2002;

21(5):490–4.

[32] Nozaki M, Kimura H, Kojima M, Ogura Y. Optical

coherence tomographic findings of the anterior seg-

ment after nonpenetrating deep sclerectomy. Am J

Ophthalmol 2002;133(6):837–9.

[33] Priglinger SG, Neubauer AS, May CA, et al. Optical

coherence tomography for the detection of laser in situ

keratomileusis in donor corneas. Cornea 2003;22(1):

46–50.

[34] Toth CA, Chiu EK, Winter KP, et al. In-vivo response

to free electron laser incision of the rabbit cornea. La-

sers Surg Med 2001;29(1):44–52.

[35] Ucakhan OO, Tello C, Liebmann JM, Ritch R, Asbell

PA. Optical coherence tomography of Intacs. J Cataract

Refract Surg 2001;27(10):1535.

[36] Ustundag C, Bahcecioglu H, Ozdamar A, Aras C, Yil-

dirim R, Ozkan S. Optical coherence tomography for

evaluation of anatomical changes in the cornea after

laser in situ keratomileusis. J Cataract Refract Surg

2000;26(10):1458–62.

[37] Wang J, Fonn D, Simpson TL, Jones L. Relation be-

tween optical coherence tomography and optical pa-

chymetry measurements of corneal swelling induced

by hypoxia. Am J Ophthalmol 2002;134(1):93–8.

[38] Wang J, Fonn D, Simpson TL, Jones L. The measure-

ment of corneal epithelial thickness in response to

hypoxia using optical coherence tomography. Am J

Ophthalmol 2002;133(3):315–9.

[39] Wong AC, Wong CC, Yuen NS, Hui SP. Correlational

study of central corneal thickness measurements on

Hong Kong Chinese using optical coherence tomogra-

phy, Orbscan and ultrasound pachymetry. Eye 2002;

16(6):715–21.

[40] Bagayev SN, Gelikonov VM, Gelikonov GV, et al.

Optical coherence tomography for in situ monitoring

of laser corneal ablation. J Biomed Opt 2002;7(4):

633–42.

[41] Bechmann M, Thiel MJ, Neubauer AS, et al. Central

corneal thickness measurement with a retinal optical

coherence tomography device versus standard ultra-

sonic pachymetry. Cornea 2001;20(1):50–4.

[42] Feng Y, Varikooty J, Simpson TL. Diurnal variation of

corneal and corneal epithelial thickness measured

using optical coherence tomography. Cornea 2001;

20(5):480–3.

[43] Hirano K, Ito Y, Suzuki T, Kojima T, Kachi S, Miyake

Y. Optical coherence tomography for the noninvasive

evaluation of the cornea. Cornea 2001;20(3):281–9.

[44] Hirano K, Kojima T, Nakamura M, Hotta Y. Triple

anterior chamber after full-thickness lamellar ke-

ratoplasty for lattice corneal dystrophy. Cornea 2001;

20(5):530–3.

[45] van den Berg TJ, Spekreijse H. Near infrared light

absorption in the human eye media. Vision Res 1997;

37(2):249–53.

[46] Laser Institute of America. American National Stand-

ard for Safe Use of Lasers. Orlando: Laser Institute of

America, American National Standards Institute, Inc.;

2000 [ANSI Z136. 1-2000].

[47] Pavlin CJ, Harasiewicz K, Sherar MD, Foster FS.

Clinical use of ultrasound biomicroscopy. Ophthal-

mology 1991;98(3):287–95.

[48] Chen HB, Kashiwagi K, Yamabayashi S, Kinoshita T,

Ou B, Tsukahara S. Anterior chamber angle biometry:

quadrant variation, age change and sex difference.

Curr Eye Res 1998;17(2):120–4.

[49] Saragoussi JJ, Othenin-Girard P, Pouliquen YJ. Ocular

damage after implantation of oversized minus power

anterior chamber intraocular lenses in myopic phakic

eyes: case reports. Refract Corneal Surg 1993;9(2):

105–9.


Anterior segment imaging: ultrasound biomicroscopy

Hiroshi Ishikawa, MD*, Joel S. Schuman, MD

UPMC Eye Center and Department of Ophthalmology, University of Pittsburgh School of Medicine, The Eye and Ear Institute,

Suite 816, 203 Lothrop Street, Pittsburgh, PA 15213, USA

High-frequency ultrasound biomicroscopy (UBM) image lines (or A-scans) at a scan rate of 8 frames

(Paradigm Medical Industries, Salt Lake City, Utah)

provides high-resolution in vivo imaging of the ante-

rior segment in a noninvasive fashion. In addition

to the tissues easily seen using conventional methods

(ie, slit lamp), such as the cornea, iris, and sclera,

structures including the ciliary body and zonules,

previously hidden from clinical observation, can be

imaged and their morphology assessed. Pathophysio-

logic changes involving anterior segment architec-

ture can be evaluated qualitatively and quantitatively.

This article discusses the role of UBM in imaging

of the anterior segment of the eye from the qualitative

and quantitative analysis point of view.

Equipment and technique

The technology for UBM, originally developed

by Pavlin, Sherar, and Foster, is based on 50- to

100-MHz transducers incorporated into a B-mode

clinical scanner [1–3]. Higher frequency transducers

provide finer resolution of more superficial structures,

whereas lower frequency transducers provide greater

depth of penetration with less resolution. The com-

mercially available units operate at 50 MHz and

provide lateral and axial physical resolutions of ap-

proximately 50 mm and 25 mm, respectively. Tissue

penetration is approximately 4 to 5 mm. The scan-

ner produces a 5 � 5 mm field with 256 vertical


doi:10.1016/j.ohc.2003.12.001

This article was supported in part by NIH contracts

RO1-EY13178 and RO1-EY11289.


E-mail address: [email protected] (H. Ishikawa).

per second.

Each A-scan is mapped into oversampled 1024

points, with 256 gray-scale levels representing the

logged amplitude of reflection, and then the number

of points is downsized to 432 pixels to fit on the UBM

monitor. The real-time image is displayed on a video

monitor and can be recorded on videotape for later

analysis. Room illumination, fixation, and accommo-

dative effort affect anterior segment anatomy and

should be held constant, particularly when quantita-

tive information is being gathered.

The image acquisition technique has been de-

scribed elsewhere and is similar to traditional immer-

sion B-scan ultrasonography [3–5]. In the Paradigm

Instruments UBM, the probe is suspended from a

gantry arm to minimize motion artifacts, and lateral

distortion is minimized by a linear scan format. In

the OTI (Ophthalmic Technologies, Toronto, Canada)

device, the probe is small and light enough not to

require a suspension arm, and a sector scanning

method is used. Scanning is performed with the

patient in the supine position. A plastic eyecup of

the appropriate size is inserted between the lids,

holding methylcellulose or normal saline coupling

medium. To maximize the detection of the reflected

signal, the transducer should be oriented so that

the scanning ultrasound beam strikes the target

surface perpendicularly.

Qualitative ultrasound biomicroscopy

The normal eye

In the normal eye, the cornea, anterior chamber,

posterior chamber, iris, ciliary body, and anterior lens

s reserved.

Fig. 1. Ultrasound biomicroscopic appearance of a normal

eye. The cornea (C), sclera (S), anterior chamber (AC), pos-

terior chamber (PC), iris (I), ciliary body (CB), lens capsule

(LC), and lens (L) can be identified. The scleral spur (black

arrow) is an important landmark to assess the morphologic

relationships among the anterior segment structures.

H. Ishikawa, J.S. Schuman / Ophthalmol Clin N Am 17 (2004) 7–208

surface can be recognized easily (Fig. 1). The scleral

spur is the only constant landmark allowing one to

interpret UBM images in terms of the morphologic

status of the anterior chamber angle and is the key for

analyzing angle pathology. The scleral spur is located

where the trabecular meshwork meets the interface

line between the sclera and ciliary body.

Generally, in the normal eye, the iris has a roughly

planar configuration with slight anterior bowing, and

the anterior chamber angle is wide and clear. Mor-

phologic relationships among the anterior segment

structures alter in response to a variety of physio-

Fig. 2. Occludable angle with dark room provocative test. (A) Th

a lighted condition. (B) The angle is completely occluded (arrows

logic stimuli (ie, accommodative targets and light);

therefore, maintaining a constant testing environ-

ment is critical for cross-sectional and longitudi-

nal comparison.

Glaucoma

Angle-closure glaucoma

Iris apposition to the trabecular meshwork is the

final common pathway of angle-closure glaucoma,

which represents a group of disorders. This condition

can be caused by one or more abnormalities in the

relative or absolute sizes or positions of anterior seg-

ment structures, or by abnormal forces in the posterior

segment that alter the anatomy of the anterior seg-

ment. Forces are generated to cause angle closure

in four anatomic sites: the iris (pupillary block), the

ciliary body (plateau iris), the lens (phacomorphic

glaucoma), and behind the iris by a combination of

various forces (malignant glaucoma and other poste-

rior pushing glaucoma types). Differentiating these

affected sites is the key to provide effective treatment.

UBM is extremely useful for achieving this goal.

Angle occludability. Examining eyes with narrow

angles requires careful attention to the occludability

of the angle. Although provocative testing, such as

dark room gonioscopy, is useful for detecting the

angle occludability, it is now rarely used, because it

is subjective, time consuming, and prone to false-

negative results owing to the difficulty of standardiz-

ing the slit-lamp light intensity. With UBM, dark room

provocative testing can be performed in a standard-

ized environment generating objective results by pro-

viding information on the state of the angle under

normal light conditions and its tendency to occlude

spontaneously under dark conditions (Fig. 2).

e anterior chamber angle is slit-like opened (arrows) under

) under a dark condition.

Fig. 4. Plateau iris. A large and anteriorly positioned ciliary

body holds the iris root up against the cornea, leading to a

partially occluded angle. The arrow represents the location

of the scleral spur.

hthalmol Clin N Am 17 (2004) 7–20 9

Pupillary block. Pupillary block is the most com-

mon type of angle-closure glaucoma. At the iridolen-

ticular contact, resistance to aqueous flow from the

posterior to the anterior chamber creates an unbal-

anced relative pressure gradient between the two

chambers, pushing the iris up toward the cornea

(Fig. 3A). This abnormal resistance causes anterior

iris bowing, angle narrowing, and acute or chronic

angle-closure glaucoma. The other anterior seg-

ment structures and their anatomic relationships re-

main normal.

Laser iridectomy equalizes the pressure gradient

between the anterior and posterior chambers and

flattens the iris. The result is a widened anterior

chamber angle (Fig. 3B).

Plateau iris. A plateau iris configuration occurs

owing to a large or anteriorly positioned ciliary

body (pars plicata), which pushes the iris root me-

chanically up against the trabecular meshwork

(Fig. 4). The iris root may be short and inserted

anteriorly on the ciliary face, creating a narrow and

crowded angle. The anterior chamber is usually of

medium depth, and the iris surface looks flat or

slightly convex, just like in a normal eye. With in-

dentation gonioscopy, the ‘‘double-hump’’ sign is ob-

served. The peripheral hump results from the rigid

presence of the ciliary body holding the iris root; the

central hump represents the center part of the iris

resting over the anterior lens surface. The space be-

tween the two humps represents the area between

the ciliary processes and the endpoint of iridolen-

ticular contact. These findings can be confirmed by

performing indentation UBM (Fig. 5), a special tech-

nique that imposes mild pressure on the peripheral

cornea with the skirt of a plastic eyecup so that one

can simulate indentation gonioscopy [6].

Phacomorphic glaucoma. Anterior subluxation of

the lens may lead to angle-closure glaucoma because

H. Ishikawa, J.S. Schuman / Op

Fig. 3. Pupillary block. (A) The angle shows appositional

closure owing to anterior bowing (arrows) of the iris. (B)

The angle is open with a flattened iris after laser peripheral

iridotomy. The patent hole on the iris (arrow) equalizes the

pressure between the anterior and posterior chambers.

of the lens pushing the iris and ciliary body toward the

trabecular meshwork.

Malignant glaucoma. Malignant glaucoma, also

known as ciliary block or aqueous misdirection,

presents the greatest diagnostic and treatment chal-

lenge. Forces posterior to the lens push the lens– iris

diaphragm forward, causing angle closure. UBM

clearly shows that all anterior segment structures are

displaced and pressed tightly against the cornea with

or without fluid in the supraciliary space (Fig. 6).

Other causes of angle closure. Iridociliary body

cysts can produce angle-closure glaucoma. The ante-

rior chamber angle is occluded partially or intermit-

tently owing to singular or multiple cysts (Fig. 7).

UBM is extremely useful in making the diagnosis

in these cases. Other entities, such as iridociliary tu-

mor, enlargement of the ciliary body owing to inflam-

mation or tumor infiltration, or air or gas bubbles after

intraocular surgery, may also present angle closure.

Open-angle glaucoma

The only type of open-angle glaucoma that shows

characteristic findings on UBM is the pigment disper-

sion syndrome. In this familial autosomal dominant

disease, mechanical friction between the posterior

iris surface and anterior zonular bundles releases iris

pigment particles into aqueous flow. These particles

are deposited on structures throughout the anterior

Fig. 6. Malignant glaucoma (composite image). The lens,

iris, and ciliary process are all pushed forward, resulting in

an extremely shallow anterior chamber and totally occluded

angle. The ciliary process (asterisk) is completely anteriorly

rotated (white arrow), probably pulled by zonules. The

scleral spur is located at the black arrow.

Fig. 7. Angle closure owing to an iridociliary cyst. An

iridociliary cyst (asterisk) pushes the iris root toward the

cornea, resulting in total occlusion of the angle (arrows).

Fig. 5. Indentation UBM on an eye with a plateau iris. The

angle is slit-like opened (the arrow represents the scleral

spur location). The ‘‘double-hump’’ sign, one hump owing

to the ciliary process (black arrow head) and the other

owing to the lens (white arrow head), is demonstrated.

(Adapted from Ishikawa H, Inazumi K, Liebmann JM, Ritch

R. Inadvertant corneal indentation can cause artifactitious

widening of the iridocorneal angle on ultrasound biomi-

croscopy. Ophthalmic Surg Lasers 2000;31(4):342–5;

with permission.)


segment. The diagnostic triad consists of a Kruken-

berg spindle, radial transillumination defects of the

midperipheral iris, and pigment deposition on the

trabecular meshwork.

Typical UBM findings associated with this con-

dition include a widely opened angle, an iris with

slight concavity (bowing posteriorly), and increased

iridolenticular contact (Fig. 8). As is true in pupillary

block, there is a relative pressure gradient between

the anterior and posterior chamber; however, because

the anterior chamber is the one that holds higher

pressure, this condition is called ‘‘reverse pupillary

block’’ [7]. Laser iridotomy eliminates this pressure

gradient, resulting in a flattened iris [8].

Abnormalities of the iris and ciliary body

Ultrasound biomicroscopy is helpful in differenti-

ating solid from cystic lesions of the iris and ciliary

body (see Fig. 7 and Fig. 9). The size of these le-

sions can be measured, and the extent to which they

invade the iris root and ciliary face can be evaluated.

In hypotony cases, UBM can distinguish tractional

from dehiscence ciliary body detachment, which re-

quires a different management approach [9].

Ocular trauma

Ocular trauma often limits the visibility of the

ocular structure owing to the presence of hyphema.

Accurate assessment of the structural damage and

locating small foreign bodies can be a challenging

task when clear direct visualization is not achieved.

UBM can be performed over a plano soft contact lens

to minimize the risk of further injury with eyecups

Fig. 8. Pigment dispersion syndrome. The angle is wide

with a concave iris (arrow). Note the extremely wide

iridolenticular contact. (Adapted from Breingan PJ, Esaki K,

Ishikawa H, et al. Iridolenticular contact decreases follow-

ing laser iridotomy for pigment dispersion syndrome. Arch

Ophthalmol 1999;117(3):325–8; with permission.)

H. Ishikawa, J.S. Schuman / Ophthalmol Clin N Am 17 (2004) 7–20 11

or with infection in a micro–open wound. With the

help of UBM, angle recession can be differentiated

plainly from cyclodialysis [10,11].

In eyes with angle recession, the ciliary body face

is torn at the iris insertion, resulting in a wide-angle

Fig. 9. Iridociliary tumor. Abnormally large ciliary process

(asterisk) involving the iris root and pars plana is visualized.

appearance with no disruption of the interface in

between the sclera and ciliary body (Fig. 10). In

contrast, in cyclodialysis, the ciliary body is detached

from its normal location at the scleral spur, creating

a direct pathway from the anterior chamber to the

supraciliary space (Fig. 11).

Foreign bodies generate various artifacts based

on their acoustic characteristics [12]. In general, ma-

terials that contain air (ie, wood and concrete) create

shadowing artifact by absorbing most of the incoming

ultrasound at their sites, whereas hard and dense

materials (ie, metal and glass) generate comet tail

artifacts by reflecting ultrasound back and forth within

the materials (Fig. 12). Scleral sutures after intraocu-

lar surgery can be identified by searching for this

shadowing artifact (by refraction) (Fig. 13).

Intraocular lens position

An intraocular lens is an easy target for UBM

visualization, because it is a type of foreign body.

Optic and haptic locations can be assessed accurately

by looking for a strong echo at their interface plane.

Because the capsular bag cannot always be visualized,

the most peripheral portion of the haptic defines its

position in the capsular bag, ciliary sulcus, or a

dislocated point (Fig. 14). This technique is used in

Fig. 10. Angle recession. Blunt trauma caused a tear into

the ciliary body face (white arrow), but the iris remained

attached to the scleral spur (black arrow). There is no direct

communication between the anterior chamber and the su-

praciliary space.

Fig. 11. Cyclodialysis. The ciliary body is avulsed from the

sclera, resulting in free aqueous flow from the anterior cham-

ber through the cleft into the supraciliary space (asterisk).

Fig. 13. Scleral suture can be identified by looking for its

shadowing artifact (arrow). This artifact is created owing to

refraction of the ultrasound beam at a boundary between

suture thread and the surrounding tissues.


various studies related to many different types of

intraocular lenses [13–17].

Quantitative ultrasound biomicroscopy

Physical resolution and measurement precision

Physical resolution is often confused with mea-

surement precision. Physical resolution specifies how

close together two objects can be located yet still

Fig. 12. Intraocular foreign body. (A) Foreign body (ar-

row head) with a material that consists of multiple cavities

inside (ie, wood and concrete) generates shadowing artifact

(arrow) by absorbing ultrasound power. The iris image is

masked by shadowing. (B) Hard and dense foreign body

(arrow head) (ie, glass and metal) creates comet tail artifact

(arrow) owing to multiple internal reflections. The iris im-

age is disrupted by the comet tail artifact. (Adapted from

Laroche D, Ishikawa H, Greenfield D, et al. Ultrasound bio-

microscopic localization and evaluation of intraocular for-

eign bodies. Acta Ophthalmol Scand 1998;76(4):491–5;

with permission.)

be determined to be distinct. It also specifies the

smallest object detectable. Measurement precision

refers to the width and height of a single pixel on

the screen that can be identified by the operator using

the screen cursor. The UBM measurement software

calculates distance and area by counting the number

of pixels along the measured line or inside the

designated area and multiplies the pixel counts by

Fig. 14. Posterior chamber intraocular lens haptic. The most

peripheral portion of the haptic is positioned within the cap-

sular bag and is located central to the ciliary process (arrow).

Table 1

Parameters proposed by Pavlin et al [1]

Name Abbreviation Description

Angle opening

distance

AOD Distance between

the trabecular

meshwork and the

iris at 500 mmanterior to the

scleral spur

Trabecular– iris angle TIA q 1 Angle of the

angle recess

Trabecular–ciliary

process distance

TCPD Distance between

the trabecular

meshwork and the

ciliary process at

500 mm anterior

to the scleral spur

Iris thickness ID1 Iris thickness at

500 mm anterior

to the scleral spur

Iris thickness ID2 Iris thickness at

2 mm from the

iris root

Iris thickness ID3 Maximum iris

thickness near the

pupillary edge

Iris–ciliary process

distance

ICPD Distance between

the iris and the

ciliary process

along the line

of TCPD

Iris–zonule distance IZD Distance between

the iris and the

zonule along the

line of TCPD

Iris– lens contact

distance

ILCD Contact distance

between the iris

and the lens

Iris– lens angle ILA q 2 Angle between

the iris and the

lens near the

pupillary edge


the theoretical size of the pixel. Measurement preci-

sion can be better than physical resolution by over-

sampling the signal.

Commercially available instruments provide lat-

eral and axial physical resolution of approximately

50 and 25 mm, respectively. The resolution of the

Paradigm device is slightly better than that of the

OTI device. The theoretical lateral and axial mea-

surement precision on the standard UBM monitor

(864 � 432 pixels) is approximately 6 and 12 mm.

Although UBM cannot distinguish two small objects

less than 25 mm apart along the axial scanning line,

it can still measure the distance between two objects

far enough apart ( > 25 mm, such as corneal thickness,

anterior chamber depth) with 12-mm precision.

Measurement accuracy

Pavlin et al [2] reported good qualitative agree-

ment of UBM images with histologic sections. Quan-

titatively, Maberly et al [18] showed good agreement

by measuring the distance from the anterior margin

of peripheral choroidal melanomas to the scleral spur

on UBM images and histologic sections.

Pierro et al [19] compared the corneal thickness

measured by UBM versus ultrasound and optical

pachymetry. The UBM measurement was similar to

the ultrasound pachymetry, whereas optical pachyme-

try showed a poor correlation with UBM and ultra-

sound pachymetry. Urbak [20] reported similar

results. Additionally, a specially prepared plastic ma-

terial was measured with UBM and scanning elec-

tron microscopy. The axial and lateral accuracies of

UBM measurements were good and reliable.

Measurement reproducibility

Tello et al [21] reported on the reproducibility of

measuring Pavlin’s parameters (described in detail

in the next section). Intraobserver reproducibility

was reasonably good, except for the angle opening

distance (AOD), but interobserver reproducibility

was not. Urbak et al [22,23] reported similar results.

Although image acquisition differences were the

major cause of this variability, the variability of the

measurement process cannot be ignored.

All of Pavlin’s parameters require multiple steps

of measurements of a distance or an angle. The pa-

rameters are measured on the UBM monitor, allow-

ing determination of a point-to-point distance or an

angle composed of two straight lines; however, this

method does not keep the previous measurement on

the same screen. It is difficult and not reproducible

to perform measurements that require multiple steps

(ie, measuring a distance along a line drawn perpen-

dicular to a line between the scleral spur and the

corneal endothelial border that is 500 mm anterior to

the scleral spur).

To minimize the variability of the measurement

process, a fully automated measurement system

would be ideal; however, with current technology,

it would be difficult to develop such a software pro-

gram. A semi-automated software system that calcu-

lates various quantitative parameters after one user

input of the reference point location is a reasonable


compromise. The UBM Pro 2000 (Paradigm Medi-

cal Industries, Salt Lake City, Utah) can measure

the AOD in a semi-automated fashion. It has dramati-

cally improved overall reproducibility (coefficient of

variation, 7.3 to 2.5; Hiroshi Ishikawa, MD, unpub-

lished data, 1998).

Fig. 15. Pavlin’s measurement parameters (see Table 1). (A) The a

line drawn from the point on the corneal endothelial surface 500 mmthe corneal endothelial surface. The trabecular– iris angle (TIA, q1)and the arms passing through the point on the meshwork 500 mm f

opposite. (B) The trabecular ciliary distance (TCPD) is defined as

and the ciliary process on the line that is perpendicular through the

the iris–ciliary process distance (ICPD). Iris thickness also can be

point near the margin (ID3). The iris–zonule distance (IZD) is def

process. The length of iris– lens contact (ILCD) and the angle at whi

easily measured.

In addition, each observer will set the reference

point on any measurement in an idiosyncratic way.

For example, when measuring corneal thickness, one

observer may tend to select a reference point slightly

more external on the epithelial surface than another

observer. This situation would result in the first

ngle opening distance (AOD) is defined as the length of the

anterior to the scleral spur to the iris surface perpendicular to

is defined as an angle formed with the apex at the iris recess

rom the scleral spur and the point on the iris perpendicularly

the distance between a point 500 mm from the scleral spur

iris. The iris thickness (ID1) is defined along this line, as is

measured 2 mm from the iris root (ID2) and at its thickest

ined as a part of theTCPD at a point just clearing the ciliary

ch the iris leaves the lens surface (iris– lens angle; ILA, q2) are

Fig. 16. Iris concavity/convexity. Iris configuration is

determined first by creating a line from the most peripheral

to the most central points of iris pigment epithelium. A

perpendicular line is then extended from this line to the iris

pigment epithelium at the point of greatest concavity or

convexity. (A) Iris convexity measurement (arrow). (B) Iris

concavity measurement (arrow).


observer measuring greater corneal thickness, assum-

ing that each observer would choose the same point as

an endothelial border. In general, repeated measure-

ment by the same observer is reasonably reproducible.

Quantitative measurement methods

Methods proposed by Pavlin and colleagues

Pavlin et al [1] established various quantitative

measurement parameters as standards (Table 1,

Fig. 15). The position of the scleral spur is used as a

reference point for most of their parameters, because

this is the only landmark that can be distinguished

consistently in the anterior chamber angle region.

Iris concavity/convexity

Potash et al [24] introduced a parameter to evaluate

the dynamic configurational change of the iris. A line

is created from the most peripheral point to the most

central point of iris pigment epithelium. A perpen-

dicular line is then extended from this line to the iris

pigment epithelium at the point of greatest concavity

or convexity (Fig. 16).

An improved method for assessing the anterior

chamber angle

There is one problem with AOD measurement,

Pavlin’s classical method of assessing the angle open-

ing, which treats the iris surface as a straight line.

Fig. 17 shows two schematics of the angle, demon-

strating exactly the same value for the AOD and the

trabecular– iris angle (TIA). Nevertheless, it is obvi-

ous that the angle on the right is gonioscopically

narrower and more likely to be occludable than the

angle on the left; therefore, irregularities of iris con-

tour and curvature need to be taken into account.

Ishikawa et al [25] defined the angle recess area

(ARA) as the triangular area bordered by the anterior

iris surface, corneal endothelium, and a line perpen-

dicular to the corneal endothelium drawn to the iris

surface from a point 750 mm anterior to the scleral

spur (Fig. 18). In this way, the iris irregularity is prop-

erly accounted for in the measurement.

The semi-automated software in the UBM Pro

2000 also calculates the ARA. After the observer

selects the scleral spur, the program automatically

processes the image, detects a border, and calculates

the ARA. The program plots consecutive AODs from

the base of the angle recess to 750 mm anterior

to the scleral spur and performs linear regression

analysis of consecutive AODs, producing two fig-

ures—the acceleration (or slope) and the y-intercept.

The acceleration describes how rapidly the angle

is getting wider, using the tangent of the angle instead

of degrees as the unit. In other words, the accelera-

tion estimates the general shape of the angle, shallow

or wide. The y-intercept refers to the distance be-

tween the scleral spur and the iris surface along the

perpendicular to the trabecular meshwork plane. This

generalized value describes the angle opening at the

level of the scleral spur. Although these parameters

may seem similar to the AOD and TIA, there is a

fundamental difference between them. Because the

acceleration and the y-intercept are purely mathe-

matical calculations based on linear regression analy-

sis of the consecutive AODs, they can be negative

numbers, which is impossible for the physically mea-

sured AOD and TIA. A negative number for the ac-

celeration means that the angle has an almost normal

configuration at its peripheral part and becomes very

shallow, or is attached to the cornea, at its central part

(ie, appositional angle closure starting at Schwalbe’s

line with space remaining in the angle recess)

(Fig. 19). A negative y-intercept means that the angle

recess is very shallow or is attached to the cornea at

its periphery, whereas it is relatively wide centrally

Fig. 17. Limitation of the conventional angle opening distance (AOD) measurement. (A) and (B) have exactly the same value

for the AOD and trabecular– iris angle (TIA, q1). Nevertheless, the angle in (B) is gonioscopically narrower and is more likely

to be occludable than the normal-appearing angle in (A).


(ie, plateau iris and synechial closure) (Fig. 20). By

using three numerical values, the ARA, the accelera-

tion, and the y-intercept, one can describe many types

of angle configuration quantitatively.

Clinical application of quantitative ultrasound

biomicroscopy analysis

Glaucoma

Anterior chamber angle parameters have been

used in various studies, such as the development of

the angle in normal infants and children in relation to

age [26], the difference between angle-closure and

normal eyes [27], and the iris convexity related to age

[28]. Ishikawa et al [25] measured the ARA, accel-

eration, and y-intercept under standardized dark and

light conditions and reported that the more posterior

the iris insertion on the ciliary face, the less likely the

provocative test would be positive. Esaki et al [29]

found that the anterior chamber angle opening in

normal Japanese eyes narrowed with age in a cross-

sectional study.

Ultrasound biomicroscopy also provides a power-

ful tool to evaluate the effect of drug instillation on

the anterior chamber angle, iris, and ciliary body.

Kobayashi et al [30] found that the angle opening

increased after the instillation of pilocarpine in eyes

with narrow angles but decreased in eyes with a wider

Fig. 18. Angle recess area (ARA). The ARA is defined as a triangular area bordered by the anterior iris surface, corneal

endothelium, and a line perpendicular to the corneal endothelium drawn from a point 750 mm anterior to the scleral spur to the

iris surface.


or normal angle. Marchini et al [31] reported that the

potent mydriatic effect of 2% ibopamine was greater

than that of 10% phenylephrine or 1% tropicamide.

Several studies have evaluated morphologic

change after surgical procedures. Marraffa et al [32]

found that loss of endothelial cells after laser irido-

Fig. 19. Negative acceleration in ARA analysis. The linear regres

that the angle almost has a normal configuration at its peripheral pa

central part (ie, the appositional angle closure began at the level o

tomy was inversely proportional to the distance of the

iridotomy from the endothelium and scleral spur.

Gazzard et al [33] reported that laser peripheral

iridotomy produced changes in iris morphology that

were different from those caused by an increase in

illumination. Chiou et al [34] measured the time

sion analysis of ARA shows negative acceleration, meaning

rt and becomes very shallow or is apposed to the cornea at its

f Schwalbe’s line).

Fig. 20. Negative y-intercept in ARA analysis. The linear regression analysis shows a negative y-intercept, indicating that the

angle recess is very shallow or is attached to the cornea at its periphery, whereas it has a relatively wide angle recess centrally

(ie, plateau iris and synechial closure).


course of the size of collagen implants after deep

sclerectomy. They quantitatively confirmed that the

collagen implant dissolved slowly within 6 to

9 months, leaving a tunnel in the sclera.

Tumor

Ultrasound biomicroscopy is effective for the di-

agnosis and management of anterior segment tumors.

Reminick et al [35] measured the size and extent of

anterior segment tumors. Marigo et al [36] described

six eyes with anterior segment implantation cysts in a

comparison of UBM images with size measurement

with histopathologic findings.

Other situations

Other ocular diseases involving the anterior seg-

ment can be assessed using UBM. Avitabile et al [37]

investigated the correlation between the thickness

at the corneal apex and disease severity in eyes with

keratoconus. Gentile et al [38] measured the ciliary

body area in uveitic eyes. Maruyama et al [39]

measured the height of ciliary detachment in eyes

with Harada disease. Trindade et al [16] studied the

relative position of the posterior chamber phakic

intraocular lens. Intraocular lens– iris touch, intra-

ocular lens–crystalline lens touch, and anterior cham-

ber shallowing were observed after implantation.

Summary

Ultrasound biomicroscopy technology has become

an indispensable tool in qualitative and quantitative

assessment of the anterior segment. Advances in soft-

ware design and algorithms will improve theoreti-

cal understanding of the pathophysiology of anterior

segment disorders. Future applications of quantita-

tive techniques will yield important information re-

garding mechanisms of angle closure, improving

understanding of the dynamic functions of the iris,

accommodation, presbyopia, and other aspects of

anterior segment physiology and pathophysiology.

References

[1] Pavlin CJ, Harasiewicz K, Foster FS. Ultrasound bio-

microscopy of anterior segment structures in normal

and glaucomatous eyes. Am J Ophthalmol 1992;113:

381–9.

[2] Pavlin CJ, Sherar MD, Foster FS. Subsurface ultra-

sound microscopic imaging of the intact eye. Oph-

thalmology 1990;97:244–50.

[3] Pavlin CJ, Harasiewicz K, SherarMD, et al. Clinical use

of ultrasound biomicroscopy. Ophthalmology 1991;98:

287–95.

[4] Tello C, Potash S, Liebmann J, et al. Soft contact lens


modification of the ocular cup for high-resolution

ultrasound biomicroscopy. Ophthalmic Surg 1993;24:

563–4.

[5] Tello C, Liebmann JM, Ritch R. An improved coupling

medium for ultrasound biomicroscopy. Ophthalmic

Surg 1994;25:410–1.

[6] Ishikawa H, Inazumi K, Liebmann JM, et al. Inad-

vertent corneal indentation can cause artifactitious

widening of the iridocorneal angle on ultrasound bio-

microscopy. Ophthalmic Surg Lasers 2000;31:342–5.

[7] Potash SD, Tello C, Liebmann J, et al. Ultrasound

biomicroscopy in pigment dispersion syndrome. Oph-

thalmology 1994;101:332–9.

[8] Breingan PJ, Esaki K, Ishikawa H, et al. Iridolenticular

contact decreases following laser iridotomy for pig-

ment dispersion syndrome. Arch Ophthalmol 1999;

117:325–8.

[9] Roters S, Engels BF, Szurman P, et al. Typical ultra-

sound biomicroscopic findings seen in ocular hypot-

ony. Ophthalmologica 2002;216:90–5.

[10] Berinstein DM, Gentile RC, Sidoti PA, et al. Ultra-

sound biomicroscopy in anterior ocular trauma. Oph-

thalmic Surg Lasers 1997;28:201–7.

[11] Park M, Kondo T. Ultrasound biomicroscopic find-

ings in a case of cyclodialysis. Ophthalmologica 1998;

212:194–7.

[12] Laroche D, Ishikawa H, Greenfield D, et al. Ultrasound

biomicroscopic localization and evaluation of intra-

ocular foreign bodies. Acta Ophthalmol Scand 1998;

76:491–5.

[13] Sewelam A, Ismail AM, El Serogy H. Ultrasound bio-

microscopy of haptic position after transscleral fixation

of posterior chamber intraocular lenses. J Cataract

Refract Surg 2001;27:1418–22.

[14] Manabe S, Oh H, Amino K, et al. Ultrasound biomi-

croscopic analysis of posterior chamber intraocular

lenses with transscleral sulcus suture. Ophthalmology

2000;107:2172–8.

[15] Landau IM, Laurell CG. Ultrasound biomicroscopy

examination of intraocular lens haptic position after

phacoemulsification with continuous curvilinear cap-

sulorhexis and extracapsular cataract extraction with

linear capsulotomy. Acta Ophthalmol Scand 1999;77:

394–6.

[16] Trindade F, Pereira F, Cronemberger S. Ultrasound

biomicroscopic imaging of posterior chamber phakic

intraocular lens. J Refract Surg 1998;14:497–503.

[17] Saragoussi JJ, Puech M, Assouline M, et al. Ultrasound

biomicroscopy of Baikoff anterior chamber phakic

intraocular lenses. J Refract Surg 1997;13:135–41.

[18] Maberly DA, Pavlin CJ, McGowan HD, et al. Ultra-

sound biomicroscopic imaging of the anterior aspect of

peripheral choroidal melanomas. Am J Ophthalmol

1997;123:506–14.

[19] Pierro L, Conforto E, Resti AG, et al. High-frequency

ultrasound biomicroscopy versus ultrasound and opti-

cal pachymetry for the measurement of corneal thick-

ness. Ophthalmologica 1998;212(Suppl 1):1–3.

[20] Urbak SF. Ultrasound biomicroscopy. III. Accuracy

and agreement of measurements. Acta Ophthalmol

Scand 1999;77:293–7.

[21] Tello C, Liebmann J, Potash SD, et al. Measurement

of ultrasound biomicroscopy images: intraobserver

and interobserver reliability. Invest Ophthalmol Vis

Sci 1994;35:3549–52.

[22] Urbak SF. Ultrasound biomicroscopy. I. Precision

of measurements. Acta Ophthalmol Scand 1998;76:

447–55.

[23] Urbak SF, Pedersen JK, Thorsen TT. Ultrasound bio-

microscopy. II. Intraobserver and interobserver repro-

ducibility of measurements. Acta Ophthalmol Scand

1998;76:546–9.

[24] Potash SD, Tello C, Liebmann J, et al. Ultrasound

biomicroscopy in pigment dispersion syndrome. Oph-

thalmology 1994;101:332–9.

[25] Ishikawa H, Esaki K, Liebmann JM, et al. Ultrasound

biomicroscopy dark room provocative testing: a quan-

titative method for estimating anterior chamber angle

width. Jpn J Ophthalmol 1999;43:526–34.

[26] Kobayashi H, Ono H, Kiryu J, et al. Ultrasound

biomicroscopic measurement of development of an-

terior chamber angle. Br J Ophthalmol 1999;83:

559–62.

[27] Marchini G, Pagliarusco A, Toscano A, et al. Ultra-

sound biomicroscopic and conventional ultraso-

nographic study of ocular dimensions in primary

angle-closure glaucoma. Ophthalmology 1998;105:

2091–8.

[28] Ochiai H, Chihara E, Chuman H, et al. Age and

increased incidence of ‘‘forward bowing’’ of the iris

in normal eyes. J Glaucoma 1998;7:408–12.

[29] Esaki K, Ishikawa H, Liebmann JM, et al. Angle recess

area decreases with age in normal Japanese. Jpn J

Ophthalmol 2000;44:46–51.

[30] Kobayashi H, Kobayashi K, Kiryu J, et al. Pilocarpine

induces an increase in the anterior chamber angular

width in eyes with narrow angles. Br J Ophthalmol

1999;83:553–8.

[31] Marchini G, Babighian S, Tosi R, et al. Comparative

study of the effects of 2% ibopamine, 10% phenyleph-

rine, and 1% tropicamide on the anterior segment.

Invest Ophthalmol Vis Sci 2003;44:281–9.

[32] Marraffa M, Marchini G, Pagliarusco A, et al. Ultra-

sound biomicroscopy and corneal endothelium in Nd:

YAG-laser iridotomy. Ophthalmic Surg Lasers 1995;

26:519–23.

[33] Gazzard G, Friedman DS, Devereux JG, et al. A pro-

spective ultrasound biomicroscopy evaluation of

changes in anterior segment morphology after laser

iridotomy in Asian eyes. Ophthalmology 2003;110:

630–8.

[34] Chiou AG, Mermoud A, Underdahl JP, et al. An ultra-

sound biomicroscopic study of eyes after deep scle-

rectomy with collagen implant. Ophthalmology 1998;

105:746–50.

[35] Reminick LR, Finger PT, Ritch R, et al. Ultrasound


biomicroscopy in the diagnosis and management of

anterior segment tumors. J Am Optom Assoc 1998;

69:575–82.

[36] Marigo FA, Finger PT, McCormick SA, et al. Anterior

segment implantation cysts: ultrasound biomicroscopy

with histopathologic correlation. Arch Ophthalmol

1998;116:1569–75.

[37] Avitabile T, Marano F, Castiglione F, et al. Kerato-

conus staging with ultrasound biomicroscopy. Oph-

thalmologica 1998;212(Suppl 1):10–2.

[38] Gentile RC, Liebmann JM, Tello C, et al. Ciliary body

enlargement and cyst formation in uveitis. Br J Oph-

thalmol 1996;80:895–9.

[39] Maruyama Y, Kimura Y, Kishi S, et al. Serous detach-

ment of the ciliary body in Harada disease. Am J Oph-

thalmol 1998;125:666–72.


Clinical applications of optical coherence tomography for

the diagnosis and management of macular diseases

Irene Voo, MD, Elias C. Mavrofrides, MD, Carmen A. Puliafito, MD, MBA*

Bascom Palmer Eye Institute and Department of Ophthalmology, University of Miami School of Medicine,

900 N.W. 17th Street, Miami, FL 33136, USA

Cross-sectional imaging of the posterior segment time-of-flight delay and intensity of back-reflected

with sufficient resolution to visualize retinal topogra-

phy and internal tissue structure in vivo has been

possible since the early 1990s owing to the develop-

ment of optical coherence tomography (OCT) [1].

This noncontact noninvasive technique uses a low-

coherence continuous-wave light source and interfer-

ometry for image formation. Differences in time delay

of the reflected light from ocular tissues are detected

by a photodiode followed by signal-processing elec-

tronics and computer data acquisition. The tomo-

graphs are digitally processed after acquisition to

compensate for micron-scale axial motion artifacts

owing to subject movement or pulsatile blood flow [2].

Optical coherence tomography provides micron-

scale axial resolution that is not limited by pupil

aperture or ocular aberrations. The temporal coher-

ence properties of the light source determine the

resolution; however, lateral eye movement, which is

not compensated for by the scan registration algo-

rithm, can degrade the image. Standard third-genera-

tion OCT systems using a superluminescent diode

light source have an axial resolution of 10 to 15 mm.

Prototype ultrahigh-resolution fourth-generation OCT

uses a femtosecond laser light source and has achieved

an axial resolution of 3 mm [3]. Reflection of light

from different tissue depths and measurement of the


doi:10.1016/j.ohc.2003.12.002



(C.A. Puliafito).

light allow characterization of anatomic structure.

The inner border of the retina and vitreous is well

defined because of the contrast of the nonreflective

vitreous and highly reflective retina. A red band of

high reflectivity from the nerve fiber layer is generally

seen. The normal fovea is identified by its character-

istic depression of the inner retinal border secondary

to the lateral displacement of tissue anterior to Henle’s

nerve fiber layer. The posterior hyaloid is occasionally

visible as a blue membrane (low reflectivity). The

posterior border of the retina is marked by a highly

reflective red layer and corresponds to the retinal

pigment epithelium (RPE) and choriocapillaris. A

dark layer of minimal reflectivity appears just anterior

to this red layer and corresponds in location and

thickness to the photoreceptors. Another dark layer

of weak reflectivity from the deep choroid and sclera

appears below the red layer of RPE and choriocapil-

laris. The intermediate layers of retina anterior to the

photoreceptors exhibit moderate reflectivity [4].

Readily available, reproducible, high-resolution

cross-sectional imaging of the retina allows diagnosis,

monitoring, and quantitative assessment of macular

pathology. OCT has become part of the routine

imaging performed at the authors’ center in patients

who have suspected or known macular diseases.

Viewing the anatomic structure of the vitreoretinal

interface, the macular layers, and the RPE and quan-

tifying these areas in microns has led to improved

clinical decision making and has altered the man-

agement of many cases when the biomicroscopic

examination and fluorescein angiogram have been

insufficient for detailed discrimination.

s reserved.

Fig. 2. Resolving stage I macular hole OS.

I. Voo et al / Ophthalmol Clin N Am 17 (2004) 21–3122

Selected case reports

Vitreoretinal interface disorders

The interface between the retina and vitreous is

well defined because of the difference in reflectivity of

the relatively acellular vitreous and the parallel-fiber

orientation of the inner retina. Disorders such as vit-

reomacular traction syndrome, epiretinal membranes,

and macular holes are readily imaged and recognized,

even by persons inexperienced in biomicroscopy.

Macular hole

Diagnosis and staging of macular holes by bio-

microscopy can be difficult for even the most experi-

enced examiners owing to simulating conditions, such

as a lamellar hole, vitreomacular traction syndrome,

and cystoid macular edema (CME) with central cyst.

Fluorescein angiography has limited utility for the

diagnosis and monitoring of macular holes. OCT

allows identification of the stage and any persistent

vitreomacular traction not evident on clinical exami-

nation, especially in the fellow eye. It is also useful for

monitoring the course of disease, whether spontane-

ous resolution or progression to a full-thickness mac-

ular hole, and the response to surgical intervention.

The likelihood of surgical success from pars plana

vitrectomy and membrane peeling can be predicted by

obtaining quantitative information about the apical

hole diameter, base diameter, and retinal edge thick-

ness by OCT. The preoperative hole diameter has

prognostic significance for postoperative success [5].

Case 1. A 69-year-old ophthalmologist presented

with a 2-month history of a ‘‘perfect circle covering

and just to the left of the fovea,’’ of which the edge

was becoming more distinct by the week. Past ocular

history was significant for laser retinopexy on the

right eye for a tear in 1999. Best-corrected visual

acuity was 20/20 OU. Confrontation fields were full to

finger-counting OU, and Amsler grid testing revealed

a small central area of metamorphopsia OS. Motility

testing was within normal limits, and there was no

afferent pupillary defect. Anterior segment examina-

Fig. 1. Stage I macular hole OS.

tion was significant for a 1+ posterior subcapsular

cataract OD and 1+ cortical changes OS not involving

the visual axis. Intraocular pressures were 21 and 19,

respectively. On dilated fundus examination, physio-

logic optic nerves, a blunted foveal reflex OS, and

laser retinopexy scars OD were noted. OCT imaging

was performed and clarified the diagnosis as an

impending macular hole, stage I OS (Fig. 1). Obser-

vation was recommended, and the visual symptoms

were resolved at follow-up examination 6 weeks later.

Repeat OCT showed the resolving macular hole as a

flatter contour of RPE detachment (Fig. 2).

Epiretinal membrane

Given the 6% incidence of epiretinal membranes

in patients aged more than 60 years and the increasing

incidence with age, this coexisting macular pathology

may limit visual outcomes in a significant proportion

of cataract patients. The membrane may have associ-

ated macular edema, significant anatomic distortion,

or even a pseudohole appearance. Diagnosis of an

epiretinal membrane through a visually significant

cataract can be difficult, but knowledge of its exis-

tence has a tremendous impact on counseling for post-

operative expectations. Furthermore, consideration

can be given to combined cataract and vitreoretinal

surgery. In addition, characterization of the epiretinal

membrane with OCT may help in preoperative plan-

ning for membrane peeling. A membrane that is

globally attached and associated with extensive retinal

edema must be approached with extreme caution.

Besides its importance in counseling regarding post-

operative expectations of visual recovery, OCT can

document the surgical response and the completeness

of epiretinal membrane removal, and can quantify a

decrease in macular thickness.

Case 2. A 67-year-old Hispanic woman presented

with a chief complaint of blurry vision, OS greater

than OD, for 5 months with difficulty reading and

seeing road signs. She had been started on Ocuvite

and dexamethasone/tobramycin (TobraDex) ophthal-

mic drops three times a day and ointment at bedtime

OU by an ophthalmologist a few days previously.

Best-corrected visual acuity was 20/25 OD and

Fig. 3. Epiretinal membrane with associated macular edema. Fig. 5. Epiretinal membrane with associated CME.

I. Voo et al / Ophthalmol Clin N Am 17 (2004) 21–31 23

20/60 OS at distance, reading Jaeger -1 OD and -7 OS

at near. Confrontation fields were full to finger-count-

ing, and Amsler grid testing revealed a large area

of central metamorphopsia OS. Ocular motility was

within normal limits, and there was no afferent pupil-

lary defect. Anterior segment examination was signifi-

cant for a 1+ nuclear sclerotic cataract OU and an

early posterior subcapsular cataract OD. Intraocular

pressures were 18 OU. On dilated fundus examina-

tion, a cup to disc ratio of 0.4 OU, an epiretinal

membrane OS, and peripheral cobblestone degenera-

tion OU were noted. There were no drusen, areas of

RPE atrophy, subretinal fluid, or hemorrhage consist-

ent with age-related macular degeneration. OCT im-

aging of the epiretinal membrane showed associated

macular edema with a central thickness of 468 mm(Fig. 3). The patient preferred to defer surgery until

the vision worsened and returned for follow-up ex-

amination 2 months later. Despite stable Snellen

visual acuity OS, she felt that the vision was much

worse, and repeat OCT showed that the central thick-

ness had increased to 536 mm (Fig. 4). Given the

patient’s worsening visual function and documented

worsening of edema by OCT, she was scheduled for

vitrectomy with membrane peeling.

Case 3. A 61-year-old Caucasian man with wors-

ening vision OD for 6 months presented for a second

opinion on the diagnosis of age-related macular de-

generation. He had a past ocular history of inferotem-

poral branch retinal artery occlusion OD 5 years

before presentation. Best-corrected visual acuity was

20/100 OD and 20/20 OS. Confrontation fields were

intact OS, with loss of the superior hemifield OD.

Amsler grid testing showed sparing of central fixation

OD. Ocular motility was within normal limits, and a

Fig. 4. Epiretinal membrane with increased macular edema

on follow-up examination.

2+ afferent pupillary defect was present OD. Anterior

segment examination was significant for an early

nuclear sclerotic cataract OU. Intraocular pressures

were 16 OU. On dilated fundus examination, an epi-

retinal membrane was noted OS. No drusen, RPE at-

rophy, subretinal fluid, or hemorrhage consistent with

age-related macular degeneration was seen. OCT im-

aging revealed a central thickness of 410 mmOD, with

an epiretinal membrane and associated CME (Fig. 5).

It was concluded that the decreasing vision OD was

secondary to the macular edema from the develop-

ment of an epiretinal membrane. The patient elected

observation and returned for follow-up examination

6 months later. Best-corrected visual acuity decreased

to 20/200 OD, and repeat OCT imaging revealed a

central thickness of 466 mm with some worsening of

the CME (Fig. 6). An injection of 0.1 mL of intra-

vitreal triamcinolone, 40 mg/mL, was given, and the

patient returned 6 weeks later. Best-corrected visual

acuity was still 20/200 OD, and OCT imaging

revealed no improvement, with a central thickness

of 465 mm and persistent CME (Fig. 7). Because

medical therapy gave no objective improvement,

surgical intervention was recommended. Quantifica-

tion of the retinal thickening was helpful in assessing

the lack of treatment response to medical therapy and

directing future therapeutic choices.

Alterations of the structural anatomy of the macula

Vitreomacular traction syndrome

Vitreous biomicroscopy is challenging, even for

the most experienced clinician. OCT allows an accu-

rate diagnosis with the ease of vitreomacular traction,

even by the relatively untrained clinician, and allows

monitoring of progression. Spontaneous resolution

Fig. 6. Epiretinal membrane with increased macular edema

on follow-up examination.

Fig. 7. Appearance on OCT imaging after intravitreal

triamcinolone for macular edema associated with epire-

tinal membrane.

Fig. 9. Persistent vitreomacular traction syndrome with

macular edema on follow-up examination.


with normalization of retinal contour or persistent

traction with progressive edema is easily documented.

OCT has become invaluable in determining the

need for and timing of surgical intervention. The post-

operative anatomic response can be correlated with

visual recovery.

Case 4. A 34-year-old black woman referred for

evaluation of panuveitis OD with questionable trac-

tion macular detachment had been treated with hourly

topical prednisolone acetate 1% (Pred Forte) for

2 months without resolution. Past ocular history was

negative for previous episodes of inflammation or

eye trauma. There were no associated systemic symp-

toms. Best-corrected visual acuity was 20/400 OD and

20/20 OS. There was no afferent pupillary defect.

Anterior segment examination was significant for

small nongranulomatous keratic precipitates, 2+ cell,

2+ flare, and pigment on the anterior lens surface OD.

The left eye did not have any signs of current or past

inflammatory episodes. Intraocular pressures were

18 and 20, respectively. On dilated fundus examina-

tion, 1+ vitreous cell and a flame hemorrhage near an

inflammatory lesion overlying the fovea with apparent

traction were present OD. Fundus examination OS

was within normal limits. OCT was performed to

better visualize the foveal lesion given the hazy view,

and to differentiate it from a macular hole. The find-

ings included vitreomacular traction with associated

macular edema and significant elevation of the neu-

rosensory retina OD (Fig. 8). OCT of the left eye

revealed a normal foveal contour, with no macular

edema or vitreomacular traction. Because the patient

Fig. 8. Vitreomacular traction syndrome with macular

edema OD.

was not responding to topical steroid therapy, oral

prednisone was begun, and the patient returned for re-

evaluation 3 weeks later. Best-corrected visual acuity

was still 20/400 OD and 20/20 OS. Anterior segment

examination was significant for no cell, trace flare

OD. Intraocular pressures were 23 and 21, respective-

ly. On dilated fundus examination, a dark gray circle

overlying the fovea OD was present, suggesting a

macular hole. OCT revealed persistent vitreomacular

traction but of a more narrow configuration (Fig. 9).

The patient continued every 2-hour topical steroid

therapy and returned in 3 weeks. Best-corrected visual

acuity remained 20/400 OD. Anterior segment exam-

ination showed an increase to 1+ cell OD. Dilated

fundus examination revealed a full-thickness macular

hole. OCT showed that it was progressing toward a

full-thickness hole OD with central fibrous membrane

formation (Fig. 10). Because of the documented pro-

gression of vitreomacular traction on OCT, surgical

intervention was recommended. The patient under-

went pars plana vitrectomy, membrane peeling, and

intraocular gas injection.

Pseudophakic cystoid macular edema

The localized intraretinal fluid accumulation of

Irvine-Gass syndrome appears on imaging as non-

reflective cystoid spaces in the outer plexiform and

inner nuclear layers. Although cystoid changes are

visible by slit-lamp biomicroscopy and fluorescein

angiography, only OCT can quantitatively assess ret-

inal thickness and demonstrate any associated RPE

structural anomalies beneath the edematous retina,

which can be obscured by leakage on angiography.

More importantly, measurements of retinal thickness

Fig. 10. Persistent vitreomacular traction syndrome with

macular edema on follow-up examination.

Fig. 12. Cystoid macular edema after cataract extraction in

a patient with idiopathic central serous chorioretinopathy

without associated RPE detachment characteristic of idio-

pathic central serous chorioretinopathy.


byOCTcorrelate more strongly with visual acuity than

the presence of leakage on angiography [6].

Case 5. A 71-year-old Caucasian woman pres-

ented with a chief complaint of ‘‘a central gray

spot and wavy lines for the past 2 months’’ OS.

Past ocular history was significant for uncomplicated

phacoemulsification and intraocular lens placement

OS 2.5 months previously and known idiopathic

central serous chorioretinopathy. The referring oph-

thalmologist had started the patient on prednisolone

acetate 1% (Pred Forte), six times a day, and diclofe-

nac (Voltaren), four times a day. Best-corrected visual

acuity was 20/20 OD and 20/30 OS. Amsler grid

testing showed a moderate area of central metamor-

phopsia OS. Intraocular pressures were 13 and 14,

respectively. Anterior segment examination was sig-

nificant for well-centered posterior chamber intra-

ocular lenses (PCIOL) OU without posterior capsular

opacification. The corneas were clear. There was no

anterior chamber cell, flare, or vitreous. Dilated fun-

dus examination revealed CME OS, peripapillary

mottling OD, and temporal macular RPE mottling

OS. Fluorescein angiography demonstrated CME in

the late frames OS (Fig. 11). OCT imaging was

performed to exclude an RPE detachment consistent

with the recurrence of idiopathic central serous

chorioretinopathy, and showed a central thickness of

505 mm, CME, and no RPE detachment (Fig. 12). The

patient was diagnosed with postcataract CME and

given an intravitreal triamcinolone injection. On fol-

Fig. 11. Late phase of fluorescein angiogram demonstrat-

ing CME.

low-up examination 2 weeks later, the best-corrected

visual acuity was 20/25 OS, with subjective improve-

ment in near-reading ability. Repeat OCT imaging

showed a central thickness of 258 mm, a return of

normal foveal contour, and no evidence of idiopathic

central serous chorioretinopathy (Fig. 13).

Diabetic retinopathy

Diabetic macular edema is a principal cause of

visual loss in diabetic patients [7]. OCT may be more

sensitive than biomicroscopy in detecting macular

edema, and can be especially useful for confirmation

in cases that are difficult to examine at the slit lamp

secondary to media opacity or cooperation [8,9].

Central macular thickness has also been shown to

correlate better with visual acuity than the presence of

leakage on fluorescein angiography [10]. The pres-

ence of persistent edema unresponsive to initial

therapy (ie, laser photocoagulation) can establish

the need for more invasive intervention (ie, intra-

vitreal triamcinolone or vitrectomy). The efficacy of

intravitreal triamcinolone for diffuse diabetic macular

edema versus traditional focal laser treatment has

been demonstrated by Martidis and associates [11].

The finding of a taut posterior hyaloid or vitreomac-

ular traction can help in selecting the therapeutic

modality and monitoring the response [12,13].

Case 6. A 69-year-old Hispanic woman presented

with a chief complaint of worsening near-vision for

Fig. 13. Resolved CME after intravitreal triamcino-

lone injection.

Fig. 14. Diabetic macular edema with cysts involving the

fovea OS.

Fig. 16. Diffuse diabetic macular edema OS.


the past year to the point at which she could no longer

read books. Past ocular history was significant for a

cataract extraction/PCIOL implantation OS 1.5 years

previously. A diagnosis of type II diabetes was made

1 year previously. Presenting vision was 20/400 OU

without correction, and best-corrected visual acuity

was 20/50 OU. Confrontation fields were full to

finger-counting OU, and motility was within nor-

mal limits. There was no afferent pupillary defect.

Anterior segment examination was significant for a

1+ nuclear sclerotic cataract, 3+ posterior subcap-

sular cataract, and 2+ cortical changes OD, and a

well-centered PCIOL OS without posterior capsular

opacification. Dilated fundus examination showed

moderate background diabetic retinopathy OU, no

clinically significant macular edema OD, and diffuse

macular edema OS. OCT imaging revealed a central

retinal thickness of 558 mm OS with CME involv-

ing the fovea (Fig. 14). Focal macular laser therapy

was performed, and the patient was lost to follow-up

for 1 year, after which she returned with a best-

corrected visual acuity of 20/400 OU. OCT imaging

confirmed the presence of diffuse macular edema

with a central retinal thickness of 405 mm OD and

459 mm OS (Figs. 15, 16). Intravitreal triamcino-

lone injection was performed OU, and the patient

returned for re-evaluation 6 weeks later. Best-cor-

rected visual acuity was still 20/400 OU, and repeat

OCT imaging showed the central retinal thickness to

be 215 mm OD and 266 mm OS, with improved but

persistent CME OU (Figs. 17, 18). The right eye now

had developed a visually significant cataract, making

examination for clinically significant macular edema

difficult. Because OCT imaging revealed that the

Fig. 15. Diffuse diabetic macular edema OD.

macular edema had improved OD, the patient was

thought to be stable from a retinal standpoint for

cataract extraction with PCIOL implantation.

Vein occlusion

Macular edema is a common complication of vein

occlusion. Although it can be detected by biomicros-

copy or fluorescein angiography, blocking defects

by hemorrhage or media opacity may obscure detec-

tion of this vision-threatening complication. OCT has

the advantage of imaging macular edema and quan-

tifying the increase in retinal thickness despite these

limiting factors. In addition to macular edema, sub-

retinal fluid accumulation and neurosensory retinal

detachment can be detected. The presence of persist-

ent edema unresponsive to initial therapy (ie, laser

photocoagulation) can establish the need for more

invasive intervention (ie, intravitreal triamcinolone).

OCT has a pivotal role in providing objective quan-

titative information to detect the presence of macular

edema that may be contributing to vision loss and to

monitor the effect of therapeutic interventions, such

as focal laser photocoagulation, intravitreal triamcin-

olone injection, adventitial sheathotomy, and pars

plana vitrectomy. It also can be helpful in differenti-

ating patients whose vision is limited by photorecep-

tor damage/retinal atrophy and not macular edema.

Case 7. An 83-year-old Caucasian woman pre-

sented with worsening vision OD for 2 weeks. Ocular

history was significant for a central retinal vein oc-

clusion OD diagnosed 2 months previously and pre-

sumed secondary to lymphoma. Best-corrected visual

Fig. 17. Improved but persistent CME OD after intravitreal

triamcinolone injection.

Fig. 18. Improved but persistent CME OS after intravitreal

triamcinolone injection.

Fig. 20. Resolved CME OD after intravitreal triamcino-

lone injection.


acuity was 20/200 OD and 20/20 OS. Confrontation

fields were full to finger-counting OU, and ocular

motility was within normal limits OU. No afferent

pupillary defect was present. Anterior segment ex-

amination showed a well-centered anterior chamber

intraocular lens OD, with a patent peripheral iridec-

tomy, PCIOL OS. No iris or angle neovascularization

was seen. Dilated fundus examination did not show

neovascularization of the disk or elsewhere. Moderate

vascular tortuosity and four quadrants of intrareti-

nal hemorrhage were present OD. OCT imaging

revealed a central thickness of 740 mm with CME

OD (Fig. 19). The patient elected to undergo intra-

vitreal triamcinolone injection, and on follow-up ex-

amination 1 month later, best-corrected visual acuity

had improved to 20/60 OD. A 2+ afferent pupillary

defect was now present, and repeat OCT testing

revealed a central thickness of 246 mm with recovery

of the normal foveal contour (Fig. 20). Three months

later, the patient returned with worsening of the best-

corrected visual acuity, again to 20/200 OD. OCT

imaging revealed a recurrence of CME, with a central

thickness of 623 mm (Fig. 21). A second intravitreal

triamcinolone injection was given. On follow-up ex-

amination 3 months later, the best-corrected visual

acuity remained 20/200 OD. Despite resolution of the

macular edema a second time, the patient’s vision

was no better, and it was concluded that retinal at-

rophy (as demonstrated by a central thickness of

179 mm and increased reflectivity of the deeper

choroidal layers) was now the limiting factor in visual

recovery (Fig. 22).

Fig. 19. Central retinal vein occlusion with associated

CME OD.

Age-related macular degeneration

The structural information provided by OCT is

becoming a valuable diagnostic adjunct to fluorescein

angiography and indocyanine green angiography for

age-related macular degeneration. Identification and

characterization of choroidal neovascular membranes

as above or below the RPE, even when poorly

visualized on fluorescein angiography, is possible.

OCT allows the detection, localization, and quantita-

tive evaluation of subretinal fluid and associated

CME. Anatomic alteration from drusenoid pigment

epithelial detachments, RPE tears, and stages of

retinal angiomatous proliferative lesions are also ef-

fectively identified with OCT. The goal of OCT is to

correlate the observed structural changes to changes

in visual acuity, not only to better understand the

mechanism of visual loss but also to predict the

benefit of various treatments. At the authors’ center,

OCT has been most useful as an adjunct to fluores-

cein angiography for monitoring the response to

treatment and for differentiating persistent edema or

subretinal fluid from fibrosis when questionable leak-

age is seen on fluorescein angiography after multiple

photodynamic therapy treatments. Subretinal fibrosis

is highly reflective on OCT and is frequently associ-

ated with overlying retinal atrophy, whereas an early

accumulation of subretinal fluid or CME associated

with the recurrence of choroidal neovascular mem-

branes appears differently on imaging.

Retinal angiomatous proliferative lesions appear

early as small intraretinal hemorrhages, often over-

lying a serous pigment epithelial detachment (stage I).

Fig. 21. Central retinal vein occlusion with recurrence of

CME 4 months after previously successful intravitreal triam-

cinolone injection.

Fig. 22. Central retinal vein occlusion with resolution of

CME after second intravitreal triamcinolone injection but

retinal atrophy. Note increased reflectivity of deeper cho-

roidal layers.

Fig. 24. Cystoid macular edema and serous pigment epi-

thelial detachment associated with choroidal neovascular

membrane OS.


As retinal angiomatous proliferative lesions evolve

into stage II, macular edema and the formation of an

intraretinal or subretinal neovascular complex arises.

Thermal laser and verteporfin therapy alone have not

been particularly useful for stage II lesions; however,

encouraging results have been observed by combin-

ing verteporfin therapy with a postlaser intravitreal

triamcinolone injection. OCT leads the clinician to

suspect and more accurately diagnose a retinal angi-

omatous proliferative lesion, because extensive CME

and RPE detachments are characteristic of a stage III

lesion. The clinician can then choose to perform a

combined verteporfin/triamcinolone treatment for

better efficacy.

Case 8. An 82-year-old Caucasian man with history

of wet age-related macular degeneration OU pre-

sented with a chief complaint of a decrease in contrast

Fig. 23. Late phase fluorescein angiogram demonstrating

leakage secondary to choroidal neovascular membrane and

associated large serous pigment epithelial detachment OS.

while reading with his better right eye. Past ocular

history was significant for thermal laser treatment

to both eyes and photodynamic therapy to the left

eye six times previously, with the most recent treat-

ment 3 months ago. Best-corrected visual acuity was

20/40 OD and 3/200 OS. Intraocular pressures were

25 OU. Anterior segment examination showed a

2+ nuclear sclerotic cataract OU. Dilated fundus ex-

amination showed subretinal fluid of the central

macula OU and a large pigment epithelial detachment

OS. Fluorescein angiography confirmed leakage sec-

ondary to choroidal neovasulcar membranes in both

eyes and a serous pigment epithelial detachment OS

(Fig. 23). OCT imaging revealed a central thickness of

355 mm OD and 320 mm OS, and CME OS much

greater than OD with an associated pigment epithelial

detachment (Fig. 24). The patient underwent photo-

dynamic therapy on both eyes and an intravitreal

Kenalog injection in the left eye only. On follow-up

examination 1 month later, best-corrected visual

Fig. 25. Late phase fluorescein angiogram with appearance

of persistent large serous pigment epithelial detachment OS.

Fig. 26. Resolution of CME and serous pigment epithelial

detachment 1 month after photodynamic therapy with in-

travitreal triamcinolone injection. Note the retinal atrophy

and increased reflectivity of deeper choroidal layers.

Fig. 28. Stage II retinal angiomatous proliferative lesion.

Note the reflective tissue above the RPE growing intra-

retinally and associated pigment epithelial detachment.


acuity was 20/50 OD and 20/400 OS. Fluorescein

angiography was repeated, and it was difficult to in-

terpret whether the serous pigment epithelial detach-

ment had resolved (Fig. 25). OCT imaging confirmed

resolution of the pigment epithelial detachment, CME,

and retinal atrophy, as demonstrated by the increased

reflectivity of the deeper choroidal layers (Fig. 26).

Case 9. A 79-year-old Caucasian woman presented

for a second opinion on the treatment for a new retinal

hemorrhage discovered on routine follow-up exami-

nation for wet age-related macular degeneration. Past

ocular history was significant for wet age-related

macular degeneration OS, with a stable pigment epi-

thelial detachment for the past 3 years. Best-corrected

visual acuity was 20/50 OD and 20/80 OS. Confron-

tation fields were full to finger-counting OU, and

Amsler grid testing showed a moderate central area

of metamorphopsia OS. Ocular motility was within

normal limits, and there was no afferent pupillary

defect. Anterior segment examination showed a 3+

brunescent nuclear sclerotic cataract OD and a well-

centered PCIOL OS without posterior capsular opac-

ity. Intraocular pressures were 19 and 16, respectively.

Dilated fundus examination showed soft drusen OD, a

small intraretinal hemorrhage with underlying subret-

inal fluid, and pigment epithelial detachment OS.

Fluorescein angiography demonstrated leakage in a

pattern interpreted as a classic subfoveal choroidal

neovascular membrane OS (Fig. 27). OCT imaging

clarified the diagnosis as a stage II retinal angioma-

tous proliferative lesion given the reflective tissue

Fig. 27. Midphase fluorescein angiogram with appearance of

classic subfoveal choroidal neovascular membrane OS.

above the RPE growing intraretinally and associated

pigment epithelial detachment (Fig. 28). The patient

underwent photodynamic therapy with intravitreal

triamcinolone injection. At 2-month follow-up exam-

ination, the best-corrected visual acuity was 20/80 OD

and 20/200 OS. Fluorescein angiography was not

easily interpreted as showing persistent or regressed

choroidal neovascular membrane (Fig. 29), and OCT

imaging was performed to clarify that the intraretinal

neovascular component had apparently regressed,

whereas the serous pigment epithelial detachment per-

sisted (Fig. 30).

Retinal detachment

Optical coherence tomography is extremely sen-

sitive in identifying neurosensory retinal elevation

because of the distinct difference in optical reflectiv-

ity between the photoreceptors and underlying RPE/

choriocapillaris. In patients with retinal detachment,

OCT can detect the presence of shallow subretinal

Fig. 29. Midphase fluorescein angiogram with appearance

of persistent classic subfoveal choroidal neovascular mem-

brane OS.

Fig. 30. Regressed intraretinal component of stage II retinal

angiomatous proliferative lesion but persistent serous pig-

ment epithelial detachment.

Fig. 32. Cystoid macular edema after successful retinal

detachment surgery by scleral buckle placement, cryother-

apy, and perfluoropropane gas injection.


fluid or CME that may be difficult to identify on

biomicroscopy, confirming a macula-involving reti-

nal detachment. OCT can also be used to distinguish

true retinal detachment from retinoschisis. Following

the repair of retinal detachment, OCT can be used to

identify the presence of shallow subretinal fluid under

the fovea or CME in patients who have delayed vi-

sual recovery despite seemingly successful surgery.

Case 10. A 34-year-old Hispanic woman was re-

ferred for surgical treatment of a macula-involving

retinal detachment. Past ocular history was significant

for high myopia (-10 OU). At presentation, best-

corrected visual acuity was 20/60 OD and 20/20 OS.

Confrontation fields showed loss of the superonasal

field OD. Ocular motility was within normal limits,

and there was no afferent pupillary defect. Anterior

segment examination showed clear lenses OU. Intra-

ocular pressures were 15 and 14, respectively. Dilated

fundus examination showed an inferotemporal retinal

detachment that extended into the macula OD. OCT

imaging to evaluate the cause of decreased vision to

the 20/60 level OD revealed a central thickness of

396 mm and CME despite foveal attachment (Fig. 31).

The patient was diagnosed as having a macula-

involving retinal detachment and underwent a scleral

buckle placement, cryotherapy, and perfluoropropane

gas injection. At the 3-month follow-up examination,

best-corrected visual acuity remained at 20/70 OD,

although the retina seemed to have undergone suc-

Fig. 31. Macula-involving retinal detachment with CME.

cessful repair. OCT imaging revealed the presence of

CME and a central thickness of 399 mm (Fig. 32), and

the patient was started on a course of topical therapy

with prednisolone acetate 1% (Pred Forte) and keto-

rolac (Acular).

Summary

Optical coherence tomography provides high-res-

olution cross-sectional images of macular pathology

in vivo. Owing to its noninvasive noncontact nature

and use of near-infrared illumination of the fundus, it

is well tolerated by patients. The images can be

obtained without dilation and are highly reproducible,

quantifying retinal thickness with an axial resolution

of 10 mm. These qualities make OCT a powerful

diagnostic tool complementary to fluorescein angiog-

raphy, photography, and biomicroscopy.

Optical coherence tomography has proved to be

particularly useful for the clinical evaluation of vit-

reoretinal interface disorders and alterations of the

structural anatomy of the macula, such as from

edema, choroidal neovascularization, and detachment

of the neurosensory retina or RPE. The information

obtained from high-resolution evaluation of retinal

anatomy allows the diagnosis of conditions that are

difficult to establish with biomicroscopy or angiog-

raphy and improves the clinician’s ability to make the

optimal treatment decision. The quantitative assess-

ment of OCT allows an objective means to monitor

disease progression and therapeutic response.

A logical application of this technology is the

evaluation of underlying macular pathology in pa-

tients considering cataract extraction. Uncovering

vitreomacular traction, epiretinal membranes, occult

choroidal neovascular membranes with minimal

CME, subretinal fluid accumulation, and RPE detach-

ments greatly impacts the clinical management of

cataract patients and the weighing of surgical risks

and benefits. OCT is a uniquely powerful means of


visualizing retinal morphology and pathology that

may not be revealed using current techniques of

biomicroscopy, fluorescein angiography, or B-scan

ultrasonography, and serves as the newest adjunct in

diagnostic technology.

References

[1] Huang D, Swanson EA, Lin CP, Schuman JS, Stinson

WG, Chang W, et al. Optical coherence tomography.

Science 1991;254(5035):1178–81.

[2] Hee MR, Izatt JA, Swanson EA, Huang D, Schuman

JS, Lin CP, et al. Optical coherence tomography of the

human retina. Arch Ophthalmol 1995;113(3):325–32.

[3] Drexler W, Sattmann H, Hermann B, Ko TH, Stur M,

Unterhuber A, et al. Enhanced visualization of macular

pathology with the use of ultrahigh-resolution optical

coherence tomography. Arch Ophthalmol 2003;121:

695–706.

[4] Puliafito CA, Hee MR, Lin CP, Reichel E, Schuman

JS, Duker JS, et al. Imaging of macular diseases with

optical coherence tomography. Ophthalmology 1995;

102(2):217–29.

[5] Ip MS, Baker BJ, Duker JS, Reichel E, Baumal CR,

Gangnon R, et al. Anatomical outcomes of surgery for

idiopathic macular hole as determined by optical coher-

ence tomography. Arch Ophthalmol 2002;120:29–35.

[6] Shahidi M, Ogura Y, Blair NP, et al. Retinal thickness

analysis for quantitative assessment of diabetic macu-

lar edema. Arch Ophthalmol 1991;109:1115–9.

[7] Moss SE, Klein R, Klein BE. The 14-year incidence

of visual loss in a diabetic population. Ophthalmology

1998;105(6):998–1003.

[8] Hee MR, Puliafito CA, Wong C, Duker JS, Reichel E,

Rutledge B, et al. Quantitative assessment of macular

edema with optical coherence tomography. Arch Oph-

thalmol 1995;113(8):1019–29.

[9] Yang CS, Cheng CY, Lee FL, Hsu WM, Liu JH. Quan-

titative assessment of retinal thickness in diabetic

patients with and without clinically significant macular

edema using optical coherence tomography. Acta Oph-

thalmol Scand 2001;79(3):266–70.

[10] Hee MR, Puliafito CA, Duker JS, Reichel E, Coker JG,

Wilkins JR, et al. Topography of diabetic macular

edema with optical coherence tomography. Ophthal-

mology 1998;105(2):360–70.

[11] Martidis A, Duker JS, Greenberg PB, Rogers AH,

Puliafito CA, Reichel E, et al. Intravitreal triamcino-

lone for refractory diabetic macular edema. Ophthal-

mology 2002;109(5):920–6.

[12] Rivellese M, George A, Sulkes D, Reichel E, Puliafito

CA. Optical coherence tomography after laser photo-

coagulation for clinically significant macular edema.

Ophthalmic Surg Lasers 2000;31(3):192–7.

[13] Lattanzio R, Brancato R, Pierro L, Bandello F, Laccher

B, Fiore T, et al. Macular thickness measured by opti-

cal coherence tomography (OCT) in diabetic patients.

Eur J Ophthalmol 2002;12(6):482–7.


Imaging in glaucoma

Daniel M. Stein, BA, Gadi Wollstein, MD, Joel S. Schuman, MD*

UPMC Eye Center, Department of Ophthalmology, University of Pittsburgh School of Medicine, The Eye and Ear Institute,

Suite 816, 203 Lothrop Street, Pittsburgh, PA 15213, USA

The detection and monitoring of glaucoma are more, numerous studies have shown that glaucoma-

multifactorial processes traditionally involving sev-

eral diagnostic modalities, including intraocular pres-

sure measurements, subjective evaluation of the

optic nerve head (ONH), and visual field testing.

These traditional methods of assessing glaucoma have

several key limitations that dictate the need for sup-

plementary approaches. Intraocular pressure is the

major identified risk factor for the development of

glaucomatous damage, and lowering intraocular pres-

sure serves to impede the progression of retinal

degenerative change [1–3]. Nevertheless, the high

interindividual variability and the diurnal variation

in the intraocular pressure have limited the use of this

parameter for the detection of the disease. Moreover,

intraocular pressure values do not indicate whether

damage has occurred, or to what extent. Visualiza-

tion of the ONH in glaucoma by ophthalmoscopic

examination or serial stereoscopic photographs is

highly dependent on observer skills, inducing high

interobserver and intraobserver variation that affects

the utility of this diagnostic modality [4,5]. Visual

field analysis through automated perimetry is a widely

used technique that is arguably the gold standard to

evaluate glaucomatous neuropathy and to monitor

disease progression. Although it is sensitive and

specific for detecting functional visual field loss,

automated perimetry has several significant limita-

tions [6–9]. The test requires the subjective input of

the tested individual; therefore, it is prone to high

short- and long-term fluctuation. This unavoidable

source of error necessitates multiple retesting to im-

prove the reliability of the technique, delaying the

recognition of glaucomatous damage [10]. Further-


doi:10.1016/S0896-1549(03)00102-0


E-mail address: [email protected] (J.S. Schuman).

tous field abnormalities may be preceded by structural

changes of the ONH [11–13] and nerve fiber layer

[14–17].

Because glaucomatous damage is largely irre-

versible, it is imperative to identify accurately eyes

with early structural changes, because they are at risk

for continued injury. It has been suggested that the

earlier glaucoma is detected and treated, the greater

the likelihood that medical or surgical intervention

will delay or prevent the progression of glaucomatous

neuropathy and subsequent functional impairment

[18–20]. This assumption underscores the need for

accurate and reproducible quantitative evaluation of

the eye. Beyond early detection, quantitative imaging

devices might be a more sensitive way to detect glau-

comatous progression when compared with clinical

qualitative assessments.

Work during the past two decades has resulted

in the development and implementation of several

imaging technologies designed to detect glaucoma-

tous neuropathy at early stages of disease. This re-

view outlines several of the most current imaging

technologies, including confocal scanning laser oph-

thalmoscopy (CSLO), scanning laser polarimetry

(SLP), optical coherence tomography (OCT), and

the retinal thickness analyzer (RTA). The technologic

underpinnings, sensitivity and specificity, and clinical

studies of each method are discussed.

Confocal scanning laser ophthalmoscopy

Device fundamentals

Confocal scanning laser ophthalmoscopy is a real-

time imaging technique that is used to produce three-

s reserved.

Fig. 1. Visual field test of patient with glaucoma (OD).

Visual field shows a superior nasal defect and questionable

superior arcuate defect. This field corresponds to the later

figures showing HRT and OCT scans of this patient.

D.M. Stein et al / Ophthalmol Clin N Am 17 (2004) 33–5234

dimensional images of the ONH. It is based on

the principle of spot illumination and spot detection

[22,23]. A spot of laser light is projected onto the

tissue, and the reflected light is detected by a sensor.

The system uses a pair of conjugated pinholes located

in front of the light source and the light detector

components. This pair ensures that only light

reflected from a defined focal plane will reach the

light detector. Light reflected from layers above or

below the focal plane is not focused to the pinhole,

leaving only a small fraction that can pass and be

detected. Out-of-focus light is highly suppressed, and

the suppression increases rapidly with increasing

distance from the focal plane. The device moves the

focal plane to acquire sequential images. Reconstruct-

ing the series of scans at the various focal planes

creates a three-dimensional topographic representa-

tion of the surface that is scanned.

The Heidelberg Retina Tomograph (HRT; Heidel-

berg Engineering, Heidelberg, Germany) is a CSLO

device. The HRT uses a diode laser beam (wave-

length, 670 nm) and captures a series of 32 sequen-

tial two-dimensional scans in a total acquisition

time of 1.6 seconds. The optical transverse resolution

of the HRT is 10 mm and the axial resolution 300 mm.

The initial scans are focused anterior to the retina;

the last scans are posterior to the bottom of the optic

cup. The system allows three options for the field of

view: 10 � 10 degrees, 15 � 15 degrees, or 20 � 20

degrees. For each focal plane, the HRT acquires

256 � 256 measuring points, recording a reflection

intensity value at each of the particular (x, y) coor-

dinates of the focal plane. The HRT registers and

aligns the 32 consecutive scans to correct for eye

movements that may occur during image acquisition.

For each (x, y) point, the machine determines the

surface location in the z-axis derived from the center

of gravity of the reflected light along the 32 scans. To

calculate absolute z-axis measurements, a ‘‘zero-

plane’’ is defined, which is determined by calculating

the mean height of the peripapillary retinal surface as

measured by a reference ring that is automatically

defined at the periphery of the scan. The data from

these measurements are color coded and presented as

a topographic and reflective map (Figs. 1–3).

The operator is required to trace the ONH margin.

Based on this contour line, the HRT operation soft-

ware automatically defines the reference plane. This

plane is located 50 mm posterior to the mean surface

height along a 6-degree arc at the inferotemporal

region of the contour line. The reference plane is

then used as a topographic cutoff. Structures below

the plane are defined as optic cup and structures

above the plane as neuroretinal rim.

Heidelberg Engineering has recently developed

a second-generation HRT machine (HRT II). This

device is essentially a small, lightweight, portable,

and automated version of the original HRT. The

transverse field of view is fixed to 15 � 15 degrees

with a 384 � 384 measuring point grid in each of

the scanned planes. The number of scanned planes

ranges from 16 to 64 depending on the optic disk

depth. This version uses an internal fixation target and

automatically acquires three consecutive series of

scans, shortening the duration of the image acquisition.

Reproducibility

Measurements with the HRT have been found to

be highly reproducible in numerous studies [24–29].

Rohrschneider et al reported a standard deviation for

each measuring point of 30 F 6 mm in glaucoma

patients and 22 F 6 mm in normal subjects [26].

The coefficient of variation (COV) of the various

HRT measurements in this study ranged from 2.9%

to 5.2% in glaucomatous eyes to 3.3% to 4.6% in

normal eyes.

Weinreb et al [30] showed that reproducibility

was improved when multiple scans were averaged

into a single data set, recommending what is now

Fig. 2. HRT scan of normal subject (OD). On the upper left image, the topographic ONH map is shown. The red area marks the

cup, and the green and blue areas mark the neuroretinal rim. The upper right image, the reflectance image, is shown with the

tracing (green) of the optic disk margin and the outline (red) of the optic cup. All of the ONH sectors are marked with the green

checkmark, signifying that they are within the normal limits based on Moorfields regression analysis. At the bottom of the

figure, a graph depicts the surface height along the contour line on the ONH margin. Note the normal ‘‘double-hump’’

appearance, with higher thickness in the superior and inferior regions.

D.M. Stein et al / Ophthalmol Clin N Am 17 (2004) 33–52 35

the clinical practice of averaging three scans. The

average standard deviation of all image elements was

35.5 mm in healthy subjects, and averaging three scans

reduced this measure to 25.7 mm. In patients with

glaucoma, the standard deviation was 40.2 mm with

one scan, 28.5 mm with three scans, and 24.1 mmwith five scans.

Measurements with the HRT in healthy volunteers

and a model eye had less variability when compared

with ONH analyzer measurements [29]. Another

study showed a significant reduction in the inter-

observer and intraobserver variability using the HRT

when compared with computer-assisted analysis of

stereoscopic photographs of the optic disk [31].

Sihota et al [32] studied test– retest variability

using the HRT II versus the first-generation HRT.

Although they found similar values, eyes with un-

corrected astigmatism of more than 1 diopter and

poor visual acuity were suggested to have higher

variability with the HRT II.

Sources of error or variability

Pupil dilation

Mikelberg et al [33] reported higher reproducibil-

ity for undilated pupils when compared with dilated

pupils. Contradictory findings were reported by

Zangwill et al [34]. In their study, dilation of the

pupil did not change most HRT parameters; how-

ever, for images of poorer quality, the differences

were more pronounced.

Contour line

The operator of the HRT must trace the margin

of the ONH to obtain the stereometric analysis.

Hatch et al [35] assessed the interobserver agreement

of HRT parameters reflecting the variation in contour

line placement. The interobserver agreement ranged

from ‘‘substantial’’ (intraclass correlation coefficient

[ICC], 0.67–0.73) for rim volume and disk area to

‘‘near perfect’’ (ICC, 0.83–0.94) for the mean height

Fig. 3. HRT scan of subject with glaucoma (OD). On the upper left image, note the extensive cupping of the ONH and the

thinnest area of neuroretinal rim in the inferotemporal region. Moorfields regression analysis marked the inferotemporal and the

nasal inferior sectors as abnormal and the temporal and the superior sectors as borderline. The typical double-hump

configuration in the lower graph was eliminated mainly owing to tissue loss at the ONH poles. These findings correspond with

the superior nasal visual field defect shown in Fig. 1 and show additional superior ONH damage that might precede the

perimetric appearance.


of contour and cup shape, suggesting that these

parameters are less dependent on the contour line

placement. Iester et al [36] devised a method using a

computer program to create the topographic map,

thereby avoiding the subjective observer’s input by

drawing the optic disk contour, and showed an in-

creased capacity to differentiate normal from glau-

comatous optic disks.

Reference plane

Many of the stereoscopic HRT measurements

are strongly related to the location of the reference

plane. Tan et al [37] showed that the positioning of

the reference plane was the most significant con-

tributor to the variability of measurements (�95%

of cases). In an effort to reduce the variation, Burk

et al [38] suggested a method of creating a ‘‘flexible’’

reference plane accounting for the subject’s particular

optic disk shape when calculating the reference plane.

Park and Caprioli [39] located the reference plane

posterior to the contour line based on nerve fiber layer

measurements as determined by OCT, and showed

an improved diagnostic capability mainly in abnormal

ONH configurations.

Blood vessels

In many cases, the HRT analyses incorporate

blood vessels into the neuroretinal rim. This inclusion

has been suggested as the reason for the higher

HRT rim measurements when compared with the

findings of optic disk photograph planimetric evalua-

tion [40,41]. Moreover, the pulsations of the retinal

arteries also cause variability in HRT measurements.

Chauhan and McCormic [42] showed that pulse

synchronization scanning resulted in a mean 13.6%

decrease in image variability.

Ability to distinguish normal from glaucomatous eyes

A high correlation has been reported between

HRT measurements and visual field global indexes

[43–45]. Numerous studies have examined the capa-


bilities of HRT in differentiating normal and glau-

comatous eyes. Cup shape measure, rim area, and cup

volume seem to be the best HRT parameters in

discriminating between eyes, although considerable

overlap exist between groups [46,47]. Improved dis-

crimination between groups has been found by com-

bining several parameters and ONH segmentation

[48–53]. Two of these methods have been incorpo-

rated into the HRT II: Mikelberg’s discriminating

analysis and Moorfields regression analysis [50,51].

When these methods are used, the specificity and

sensitivity to differentiate between normal subjects

and patients with early glaucoma reach 78% and

89%, respectively, with Mikelberg’s discriminating

analysis and 96.3% and 84.3%, respectively, with

Moorfields regression analysis [50,51]. Implement-

ing more sophisticated mathematical approaches

such as neural networks might provide further im-

provement [54].

The capability of the HRT to detect ONH glau-

comatous changes before the appearance of a visual

field defect was evaluated by Kamal et al [55] who

examined a small group of ocular hypertensive indi-

viduals in whom reproducible visual field changes

subsequently developed. Various parameters were

found to be significantly different from those of nor-

mal subjects.

Chauhan et al [56] have described a new statisti-

cal technique for detecting changes in topography

over time with HRT using an analysis of variance

technique. When this technique is used, a higher

frequency of progression is detected by the HRT

when compared with visual field progression, which

might suggest higher sensitivity of this method for

longitudinal evaluation; however, the specificity re-

mains uncertain [57].

Strengths and limitations

Strengths

� Rapid simple operation of the device (HRT II)� Three-dimensional, topographic representation

of the ONH� No pupil dilation necessary� Advanced data analysis capability built-in to

machine

Limitations

� The use of a reference plane is required.� Manual tracing of the ONH margin must be

performed by the operator.� Measurements may be affected by blood vessels.

� The CSLO method is appropriate for scanning

the ONH only and may be affected by anatomic

variation or pathologic processes. Although soft-

ware exists for nerve fiber layer and macular

analysis, the physical limitations of axial reso-

lution for CSLO for these purposes preclude

useful quantitative thickness measurements with

this device in these areas.

Scanning laser polarimetry

Device fundamentals

Scanning laser polarimetry is an imaging technol-

ogy optimized for quantitative measurements of the

nerve fiber layer for the detection and evaluation of

glaucoma. It uses the birefringent properties of the

retinal nerve fiber layer to quantify its thickness. The

parallel arrangement of the microtubules within

the retinal ganglion cell axons causes a quantifiable

change in the polarization of light that passes through

them. This change is called retardation, and its

numerical value is proportionate to the thickness of

the nerve fiber layer [58].

Currently, SLP is in its fourth commercial gen-

eration (Nerve Fiber Analyzer [NFA], NFA II, GDx,

GDx VCC; Laser Diagnostic Technologies, San

Diego, California). The GDx device uses a diode

laser (wavelength, 780 nm) for the illumination and

measurement of a 15 � 15 degree area of retina

with an acquisition time of 0.7 seconds. Pupil dilation

is not necessary, because studies have shown that

changing the pupil size does not significantly affect

SLP measurements [59,60]. Data from the scanned

area are displayed as a 256 � 256 pixel color-coded

grid signifying different levels of retardation and

nerve fiber layer thickness. The GDx provides nu-

merous quantitative parameters, all of which are

calculated from a band 10 pixels wide in an ellipse

concentric to the ONH (Figs. 4–6).

In an early validation study, Weinreb et al [61]

correlated the histopathologic findings in fixed mon-

key eyes with retardation measurements and demon-

strated a good correlation (r = 0.83) between

retardation and retinal nerve fiber layer thickness.

Because mainly the cornea and, to a lesser degree,

the lens and the vitreous have birefringent properties

in addition to those of the nerve fiber layer, a com-

pensator accounting for these components has been

incorporated into the GDx. Before the development

of the current version of the GDx variable corneal

compensator (VCC), the SLP device used a fixed

compensator, assuming that all eyes had a slow axis

Fig. 4. GDx VCC scan of normal subject (OD). This figure shows the fundus image (top left), a pseudocolor representation of the

nerve fiber layer thickness (middle left), and a map of areas that deviate from normal thickness values (bottom left). The nerve

fiber layer graph shows measured thickness values (green line) superimposed on the normative range (green band).


of corneal birefringence of 15 degrees in the nasal and

downward direction with a magnitude of 60 nm.

Studies by Greenfield et al [62] and Knighton et al

[63] revealed a large variation in polarization prop-

erties of the cornea in different subjects. This varia-

tion should be compensated properly to obtain

accurate nerve fiber layer measurements. This ob-

servation led to the development of the GDx VCC,

which allows for customized compensation based

on the individual corneal properties. The VCC uses

the radial birefringence of Henle’s fiber layer in the

macula as a control for measurement of corneal

birefringence. Based on these measurements, the

VCC is adjusted for each eye.

Many of the studies conducted with the GDx must

be re-evaluated using the GDx VCC before the utility

of this technology can be evaluated in the diagnosis

and management of glaucoma. The information

presented in following section is discussed mainly

in terms of its applicability to the GDx VCC system.

Reproducibility

Reproducibility data are sparse for the GDx

VCC; however, it can be assumed that this device

will demonstrate improved reproducibility when com-

pared with the original GDx. Weinreb et al calculated

the reproducibility of their early SLP system as the

Fig. 5. Visual field test of subject with glaucoma (OD).

Visual field shows an inferior hemifield defect. This field

corresponds to the following figure showing a GDx VCC

scan of a glaucomatous subject.

D.M. Stein et al / Ophthalmol C

mean standard deviation of each pixel in the mean

images. They found that the overall mean standard

deviation was 1.0 degree [58]. Hoh et al [64] found

an intraoperator COV for the NFA II device for nerve

fiber layer thickness measurements of 4.5% and 4.9%

for two operators, with decreased reproducibility

between operators. Zangwill et al [65] reported a

mean COV for interoperator reproducibility for the

NFA II of 4.2%. Hollo et al found the reproducibility

to be comparable in normal and glaucoma groups,

with a COV between 3% and 8.9% for the different

sectors investigated [60]. Waldock et al [66] also

reported high interoperator reproducibility of SLP

retardation values and demonstrated improvement

of reproducibility with the use of a blood vessel

removal algorithm.


Ethnicity and age

Ethnicity and age have an effect on retinal nerve

fiber layer thickness measurements by SLP, and these

population factors must be taken into consideration

when interpreting results [67,68].

Corneal surgeries

Owing to the corneal birefringence properties,

surgical procedures in the cornea (eg, laser-assisted

in situ keratomileusis [LASIK], corneal transplanta-

tion) can induce erroneous measurements. The intro-

duction of subject-specific corneal compensation is

likely to resolve this source of variability.

Vitreous opacities

Pons et al [69] studied the effect of vitreous

opacities on retardation measurements and showed

that artifacts produced by these irregularities could

falsely increase the value obtained for mean retinal

nerve fiber layer thickness.

Motion artifacts

The possibility that motion of the eye during

scanning could interfere with measurements was

investigated by Colen and Lemij [70]. Motion arti-

facts led to an increase in retardation and affected

several GDx parameters. This increase was shown to

be highly variable, and caution was recommended

when interpreting images with such artifacts.

Split nerve fiber layer bundles

The GDx images of the peripapillary nerve fiber

layer allow the visualization of nerve fiber bundles.

A study by Colen and Lemij showed that split

nerve fiber layer bundles—areas in the GDx image

where higher retardation is clearly divided into two

symmetric-appearing parts—are a common SLP find-

ing in healthy eyes, especially in the superior re-

gion [71]. It is likely that these areas will continue

to be seen with the use of GDx VCC. This finding

can be considered normal in patients who have glau-

coma even though it can influence some of the built-

in GDx parameters, such as the superior maximum

or symmetry.

Macular pathologies

Bagga et al [72] noted that anterior chamber bire-

fringence compensation in GDx VCC is based on

measurements of the macula as an intraocular polar-

imeter and is highly dependent on an intact Henle’s

layer. This dependence might induce a source of er-

ror in measurement, especially in instances of macu-

lar structural disease.


Numerous studies have assessed the ability of the

GDx to distinguish normal from glaucomatous eyes

[73–86]. Nevertheless, because of the inappropriate

fixed corneal compensation in many cases using

previous GDx versions, the estimates of the sensitiv-

ity and specificity of this diagnostic modality are of

questionable utility.

In a study that compared the older fixed corneal

compensator GDx system with the modified GDx

VCC system, Weinreb et al [87] showed that the

VCC device had an increased sensitivity and in-

creased areas under the receiver operator characteris-

lin N Am 17 (2004) 33–52 39

Fig. 6. GDx VCC scan of subject with glaucoma (OD). Note the blue shaded area in the superior region of the thickness map

that signifies a relatively thin nerve fiber layer. This thinning of the superior region is highlighted in the deviation map and

the nerve fiber layer graph showing thickness values outside of the norm in the superior region. This defect corresponds well

with this subject’s visual field as shown in Fig. 5.


tic (ROC) curve, a measure of the overall dis-

criminating ability of a test that incorporates sensi-

tivity and specificity. The areas under the ROC for

the various GDx VCC parameters ranged from 0.75

to 0.83 versus 0.62 to 0.68 without the corneal

compensation. Similar findings in other studies con-

firm that the GDx VCC device has a better correlation

with visual function than the original GDx with fixed

compensation [88,89]. Another study showed that

adding individualized compensation to GDx scans

allowed the investigators to obtain retardation maps

that demonstrated defects that matched those ob-

served in red-free fundus photographs, whereas scans

with fixed compensation did not correlate [90].

One of the most promising GDx parameters in

terms of identifying glaucomatous changes is a neural

network product called ‘‘the number’’ [91–93]. Be-

cause this neural network was developed using

GDx data, it cannot be applied to GDx VCC data.

Studies are underway to determine the new number

using GDx VCC data.

Medeiros et al [94] applied Fourier analysis to the

retinal nerve fiber layer thickness GDx VCC mea-

surements to improve the ability of this device to


distinguish normal versus glaucomatous eyes. Apply-

ing the Fourier-based linear discriminant function

gave a sensitivity of 84% for a given specificity of

92%. This result was a significant improvement over

the parameters given automatically by the device.


Strengths

� Rapid simple operation of device� No pupil dilation necessary

Limitations

� Limited data are available regarding the VCC

system.

Fig. 7. OCT retinal nerve fiber layer thickness of a normal subject

displays 256 adjacent A-scans. The uppermost red layer delineated w

a graph shows the patient’s nerve fiber layer thickness values (bla

charts show average nerve fiber layer thickness values for the depi

the normative database.

� The SLP-VCC method requires an internal ref-

erence polarimeter (macula region), which might

be affected by macular pathologies.� The SLP-VCC method can be used to scan the

peripapillary region only and may be affected

by anatomic variations or pathologic processes.

Optical coherence tomography

Device fundamentals

Optical coherence tomography is a noncontact

noninvasive imaging technology that uses light to

create high-resolution, real-time, cross-sectional to-

mographic images [95]. OCT is the optical equivalent

. At the top of the figure, a color-coded cross-sectional map

ith the white lines is the nerve fiber layer. Below this image,

ck line) in comparison with normative values. The circular

cted regions, and the colors denote the correspondence with


of B-scan ultrasonography wherein light reflection

from the scanned area is detected. The machine can

differentiate layers in the retina owing to differences

in the time delay of reflections from various compo-

nents of the tissue.

The OCT device uses a light source consisting

of a near-infrared, low-coherence superluminescent

diode laser (wavelength, 850 nm) split at a 50/50

coupler into two arms. One arm sends light to the

actual sample, and the other sends light to a reference

mirror. For the two beams to recombine, their pulses

must arrive at nearly the same time. An interference

signal is created when path lengths of the reference

and measurement arms are closely matched to within

the coherence length of the light. A series of these

scans is created by panning the light source across

Fig. 8. OCT retinal nerve fiber layer thickness of a subject with g

nerve fiber layer when compared with the normal thickness. The th

corresponding to the visual field defect shown in Fig. 1. Abnor

reflecting structural damage not yet evident by perimetry.

the sample, which ultimately results in a two-dimen-

sional, color-coded map based on detection of the

previously described interference signals. The device

can scan the macula, peripapillary, and ONH regions

(see Fig. 1 and Figs. 7–11). The peripapillary scan is

a circular scan optimally taken with a diameter of

3.4 mm centered at the ONH [96]. The macular

and ONH scans are composed from six linear scans

in a spoke pattern configuration equally spaced

30 degrees apart. The machine automatically identi-

fies the ONH margin as the endings of the retinal

pigment epithelium layer, eliminating the need for

subjective definition of the margin by the operator.

Although it has been recommended that the images

be acquired after pupil dilation, satisfactory scans can

be achieved in most cases without dilation.

laucoma. The cross-sectional map shows diminution of the

inning is pronounced for the most part in the inferior sectors

mal thinning is also evident in the upper sectors, possibly

Fig. 9. OCT macular map of a normal subject (OS). The uppermost image of the figure is a color-coded, cross-sectional map

along the vertical line, one of the six radial scans that compose the macular map. The central thinning corresponds to the foveola,

and the white lines mark the vitreoretinal and the retinal pigment epithelium boundaries of the retina. The lower left map is the

color-coded macular thickness map wherein blue signifies thinner retina and yellow-green thicker retina. The center map gives

the quantitative measurements in nine sectors.


Several generations of the commercial version

of the OCT device (Carl Zeiss Meditec, Dublin,

California) have been developed. The first-genera-

tion OCT 1 has transverse and axial resolutions of

approximately 20 mm and 10 to 15 mm, respectively.

The second-generation OCT 2 has similar hardware

with an improved user interface. Both generations

acquire 100 vertical A-scans in a standard OCT scan

in an acquisition time of approximately 1.2 seconds.

The recently released third-generation OCT 3 ma-

chine has improved axial resolution of 8 to10 mm.

The sampling densities can be selected by the user as

512, 256, or 128 vertical A-scans.

An experimental ultrahigh-resolution OCT system

has been developed using a Ti:Al2O3 laser that

provides an improved axial resolution of 2 to 3 mm[97]. This resolution makes it possible to identify

otherwise unseen intermediate retinal layers, such

as the retinal ganglion cell layer.

Reproducibility

Several studies have confirmed the high repro-

ducibility of OCT measurements [96,98–102]. Using

a prototype OCT device, Schuman et al found that the

standard deviation of nerve fiber layer thickness

measurements was 10 to 20 mm for the mean overall

and 15 to 30 mm for clock-hour measurements [96].

Jones et al [103] calculated the mean COVs for reti-

nal nerve fiber layer data for global thickness, quad-

rant measurements, and 12-segment subdivisions and

found them to be 5%, 8%, and 9%, respectively.

Carpineto et al [104] showed reproducible nerve fi-

ber layer thickness measurements in normal and

Fig. 10. OCT macular map of a subject with glaucoma (OD). The uppermost image shows a marked diminution of the upper red

band, signifying damage to the nerve fiber layer. The color-coded map (lower left) demonstrates a thinning of the macula, most

extensively in the inferotemporal region, which corresponds to the visual field defect shown in Fig. 1. This finding also can be

appreciated from the numerical values in the lower center map.


glaucomatous eyes using various circumpapillary

scan radii and attributed the greatest variability to

intersubject differences. Other investigators have

shown that the reproducibility of OCT measurements

is significantly increased with increased sampling

density [105].


Age

From the earliest studies with OCT [106] to

more recent studies [107,108], it has been shown that

nerve fiber layer thickness decreases with age; there-

fore, it has been suggested that the OCT normative

values should be developed in a manner that is ad-

justed for this aging effect.

Sampling density

Although OCT offers the highest axial resolu-

tion available for ophthalmic imaging, the transverse

sampling density is limited in all three scan regions.

Using the OCT 1 and 2 devices, nerve fiber layer

measurements are made by scanning a 360-degree

peripapillary region, but the circle is sampled for

only 100 data points (25 points per quadrant) along

this path. Gurses-Ozden et al have shown that

increasing sampling density along this path (to

100 points per quadrant) decreases variability in

nerve fiber layer thickness measurements [105].

Other investigators studying the reproducibility of

OCT measurements have also suggested the limited

number of sampling points as a source of measure-

ment variability [98]. The OCT 3 device allows the

user to select up to 512 sampling points, which is

likely to address this issue. Furthermore, as described

previously, the macular and ONH scans are com-

posed of linear scans in a spoke pattern configuration

with interpolation of the data for the areas between

scans. It is likely that the unscanned areas between

the linear spokes and the unsampled points along

Fig. 11. (A) OCT analysis of the ONH of a normal subject (OS). The image on the left is a inferonasal to superotemporal optical

cross-section map, one of the six radial scans that compose the ONH map. The edge of the retinal pigment epithelium/

choriocapillaris layer is marked by the blue cross, and a straight line connects between the margins. A parallel line located

anteriorly to this line separates the rim (above the line) and the cup (below the line). On the right, the contours of the ONH (red

circle) and optic cup (green circle) are displayed as created from the data obtained from all six radial scans (blue lines; yellow

line represents the scan depicted in the image on the left). (B) OCT analysis of the ONH of a subject with glaucoma (OD). Note

the widened optic cup and increased slope of the contour of the rim in the image on the left, representing axonal loss in the optic

cup of a glaucomatous patient. The contour map on the right demonstrates a large cup, with thinning of the neuroretinal rim most

pronounced in the inferotemporal region, which corresponds to the visual field defects in this patient (see Fig. 1).


each spoke also contribute to variation in OCT mea-

surements [109].


Numerous studies have evaluated the capability

of OCT measurements to distinguish between normal

and glaucomatous eyes. Schuman et al showed that

OCT nerve fiber layer measurements were signifi-

cantly thinner in glaucomatous eyes when compared

with normal eyes [106]. This observation was con-

firmed by Bowd et al [110] who found that the

mean overall nerve fiber layer and the layer in

quadrants were thinner in glaucomatous versus nor-

mal and ocular hypertensive eyes. Soliman et al [111]

showed that retinal nerve fiber layer measurements on

OCT had a higher diagnostic accuracy when com-

pared with standard red-free photographic evaluation

of glaucomatous eyes. Bowd et al found that the

largest area under the ROC curve for distinguishing

between normal and early glaucomatous eyes was for

OCT inferior quadrant thickness (0.91) [78].

Recent studies have investigated the OCT macu-

lar and ONH capabilities in differentiating between


groups. Greenfield et al [112] found a significant

correlation between macular thickness and global

visual field indexes. Guedes et al reported a signifi-

cant reduction in macular thickness in glaucoma-

affected eyes when compared with normal eyes. The

area under the ROC curve for discriminating between

normal and advanced glaucomatous eyes was 0.80

in this study. A higher area under the ROC curve was

found in the same study for circumpapillary nerve

fiber layer measurements [109]. Lederer et al [113]

used retinal macular volume to demonstrate a signifi-

cant difference between normal and early glaucoma-

tous eyes. Schuman et al [114] found that OCT

analysis of the ONH provided useful parameters for

differentiating between groups, with an area under

the ROC curve of 0.79.

A recent study by Essock et al [115] showed that

by applying Fourier-based linear discriminant analy-

sis to OCT nerve fiber layer data (OCT 2), an

increased area under the ROC curve could be dem-

onstrated (0.92) when compared with the results

using other parameters such as mean nerve fiber layer

thickness (0.87) or inferior thickness (0.89).


Strengths

� Provides a cross-sectional view of examined

tissue� Highest axial resolution� Multiple scanning regions� Automatic definition of ONH margin

Limitations

� Limited transverse sampling

Retinal thickness analyzer

Device fundamentals

Use of the RTA is similar to slit lamp biomicros-

copy as it projects a thin laser slit obliquely onto the

retina. Owing to the oblique projection, the back-

reflected light contains two light peaks correspond-

ing to the vitreoretinal interface and the chorioretinal

interface. The distance between these layers reflects

the retinal thickness [116,117]. The commercial

version of this technology (Talia Technology, Neve

Ilan, Israel) uses a green HeNe laser (540 nm) that,

when projected onto the fundus, is 3 mm in length

and approximately 10 mm in width. The RTA acquires

16 optical cross-sections approximately 100 mm apart,

covering a total area of 3 � 3 mm in 400 msec.

Five of these scans are obtained for coverage of

20 � 20 degrees (6 � 6 mm) around the fovea using

internal fixation targets to direct the eye to the proper

position for each scan. Similarly, four scans are

acquired for imaging of the peripapillary region and

four scans for the ONH. The total acquisition time for

the 13 scans is approximately 10 minutes. The axial

length and refractive error of the examined eye are

measured for automatic calibration of the measured

thickness. Because the incoming light is projected at

an angle and the outgoing light is distant from the

incoming light at the pupil level, pupil dilation of

no less than 5 mm is required for obtaining images.

Zeimer et al reported that the depth precision

of the device—the ability to determine accurately

the location of a single interface—was 5 to 10 mm.

The depth resolution of the device—the minimal

distance between two surfaces that can be resolved

by the device—was 50 mm [117].

Reproducibility

Zeimer et al used a prototype instrument and

demonstrated good reproducibility, with a standard

deviation for the measured data of F9 mm for in

vitro measurements [116]. In a later in vivo study,

they showed an intravisit COV of 3.6% (11.5 mm),

a single-scan intervisit reproducibility of 4.1%

(13.1 mm), and a three-scan average intervisit repro-

ducibility of 3.2% (10.2 mm) [117]. Other studies

have found that the RTA reproducibly measures

foveal thickness and demonstrates high interexaminer

reproducibility [100,118,119].


Media opacities

The RTA is particularly susceptible to media

opacities. In a study by Polito et al [120], a large

number of RTA measurements were unusable sec-

ondary to interference from media opacities.

Focusing images

In a study of foveal thickness measurements in

patients with normal and edematous retinas, Neuba-

uer et al compared RTA measurements with OCT

measurements. Although measurements using the

RTA were reproducible, they were relatively higher

when compared with OCT measurements [118].

The investigators suggested that the overestimation

might be related to ‘‘fuzzy’’ images despite adequate

focusing according to the user manual.

ol Clin N Am 17 (2004) 33–52 47


In a pilot study by Zeimer et al [121], quantitative

retinal thickness losses were detected in the poste-

rior pole of patients who had glaucoma.


Strengths

� Multiple scanning regions� Limited cross-sectional view

Limitations

� Pupil dilation required� Highly affected by media opacities

D.M. Stein et al / Ophthalm

Studies involving multiple imaging technologies

Confocal scanning laser ophthalmoscopy and optical

coherence tomography

Device agreement

Mistlberger et al [122] compared OCT measure-

ments with HRT topographic parameters and found

a significant correlation between mean OCT nerve

fiber layer thickness measurements and five HRT

parameters: rim area, cup–disk ratio, cup shape

measure, nerve fiber layer thickness, and nerve fiber

layer cross-sectional area (P<0.02). Schuman et al

looked at the relationship of HRT and OCT with

respect to ONH measurements [114]. HRT and OCT

measurements of disk area, cup–disk ratio, cup area,

and cup volume were highly correlated (r = 0.75–

0.85, P<0.0001), whereas the rim volume was mod-

erately correlated (r = 0.54, P<0.0001).

Sensitivity and specificity comparisons

In a comparative cross-sectional study by Sanchez

Galeana et al, HRT optic disk scans and OCT nerve

fiber layer analysis were compared to determine their

sensitivity and specificity in discriminating normal

versus early to moderate glaucomatous eyes [82].

HRT sensitivity ranged from 64% to 75% for the

various parameters, and specificity ranged from

68% to 80%. OCT achieved a sensitivity between

76% and 79% and a specificity of 68% to 81%. Zang-

will et al compared the ability of HRT and OCT to

distinguish normal and glaucomatous eyes using area

under the ROC curves and found no significant differ-

ences between the best discriminating parameter from

each instrument [83]. Schuman et al reported areas

under the ROC curve for an association with clinical

diagnosis among OCT and HRT measurements of the

ONH [114]. The two devices demonstrated similar

performances, with an area under the ROC ranging

from 0.47 to 0.79 for the various parameters.

Optical coherence tomography and scanning laser

polarimetry

Device agreement

A recent study by Bagga et al demonstrated a

significant correlation between several SLP (GDx

VCC) parameters and OCT-derived nerve fiber layer

thickness measurements [89]. This correspondence

was shown to be stronger with VCC than with the

fixed compensation model.

Optical coherence tomography and retinal thickness

analysis

Device agreement

Polito et al [21] reported excellent agreements

between OCT and RTA measurements of foveal

retinal thickness, although they found that media

opacities created less interference in OCT images

than in RTA. In contrast, another study showed that

RTA measurements of foveal thickness differed

from OCT measurements in that they were consist-

ently higher, a finding attributed to the different

definition of the measured thickness between the two

devices [118].

Summary

Structural assessment using the imaging technolo-

gies discussed herein provides reproducible quantita-

tive measurements of posterior segment ocular

structures. These measurements have been found to

provide useful data for glaucoma detection in various

regions of the posterior segment. Further studies are

needed to evaluate the utility of these technologies

for pre-perimetric glaucoma detection and for moni-

toring glaucoma progression over an extended period.

References

[1] Leske MC, Heijl A, Hussein M, et al. Factors for

glaucoma progression and the effect of treatment:

the early manifest glaucoma trial. Arch Ophthalmol

2003;121(1):48–56.

[2] Goldberg I. Relationship between intraocular pressure


and preservation of visual field in glaucoma. Surv

Ophthalmol 2003;48:S3–7.

[3] The Advanced Glaucoma Intervention Study (AGIS).

7. The relationship between control of intraocular

pressure and visual field deterioration. The AGIS In-

vestigators. Am J Ophthalmol 2000;130(4):429–40.

[4] Lichter PR. Variability of expert observers in eval-

uating the optic disc. Trans Am Ophthalmol Soc

1976;74:532–72.

[5] Gaasterland DE, Blackwell B, Dally LG, et al. The

Advanced Glaucoma Intervention Study (AGIS): 10.

Variability among academic glaucoma subspecialists

in assessing optic disc notching. Trans Am Ophthal-

mol Soc 2001;99:177–84.

[6] Burnstein Y, Ellish NJ, Magbalon M, Higginbotham

EJ. Comparison of frequency doubling perimetry

with Humphrey visual field analysis in a glaucoma

practice. Am J Ophthalmol 2000;129(3):328–33.

[7] Sekhar GC, Naduvilath TJ, Lakkai M, et al. Sensitiv-

ity of Swedish interactive threshold algorithm com-

pared with standard full threshold algorithm in

Humphrey visual field testing. Ophthalmology 2000;

107(7):1303–8.

[8] Wadood AC, Azuara-Blanco A, Aspinall P, et al. Sen-

sitivity and specificity of frequency-doubling technol-

ogy, tendency-oriented perimetry, and Humphrey

Swedish interactive threshold algorithm-fast perime-

try in a glaucoma practice. Am J Ophthalmol 2002;

133(3):327–32.

[9] Johnson CA, Sample PA, Cioffi GA, et al. Structure

and function evaluation (SAFE). I. Criteria for glau-

comatous visual field loss using standard automated

perimetry (SAP) and short wavelength automated

perimetry (SWAP). Am J Ophthalmol 2002;134(2):

177–85.

[10] Keltner JL, Johnson CA, Quigg JM, et al. Confirma-

tion of visual field abnormalities in the Ocular

Hypertension Treatment Study. Ocular Hypertension

Treatment Study Group. Arch Ophthalmol 2000;

118(9):1187–94.

[11] Sommer A, Pollack I, Maumenee AE. Optic disc pa-

rameters and onset of glaucomatous field loss. I.

Methods and progressive changes in disc morphol-

ogy. Arch Ophthalmol 1979;97(8):1444–8.

[12] Pederson J, Anderson D. The mode of progressive

disc cupping in ocular hypertension and glaucoma.

Arch Ophthalmol 1980;98:490–5.

[13] Quigley HA, Katz J, Derick RJ, et al. An evaluation

of optic disc and nerve fiber layer examinations in

monitoring progression of early glaucoma damage.

Ophthalmology 1992;99(1):19–28.

[14] Sommer A, Miller NR, Pollack I, et al. The nerve

fiber layer in the diagnosis of glaucoma. Arch Oph-

thalmol 1977;95(12):2149–56.

[15] Sommer A, Quigley HA, Robin AL, et al. Evaluation

of nerve fiber layer assessment. Arch Ophthalmol

1984;102(12):1766–71.

[16] Sommer A, Katz J, Quigley HA, et al. Clinically

detectable nerve fiber atrophy precedes the onset of

glaucomatous field loss. Arch Ophthalmol 1991;

109(1):77–83.

[17] Quigley HA, Addicks EM, Green WR. Optic nerve

damage in human glaucoma. III. Quantitative correla-

tion of nerve fiber loss and visual field defect in glau-

coma, ischemic neuropathy, papilledema, and toxic

neuropathy. Arch Ophthalmol 1982;100(1):135–46.

[18] Hattenhauer MG, Johnson DH, Ing HH, et al. The

probability of blindness from open-angle glaucoma.

Ophthalmology 1998;105(11):2099–104.

[19] Chen PP. Blindness in patients with treated open-angle

glaucoma. Ophthalmology 2003;110(4):726–33.

[20] Oliver JE, Hattenhauer MG, Herman D, et al. Blind-

ness and glaucoma: a comparison of patients pro-

gressing to blindness from glaucoma with patients

maintaining vision. Am J Ophthalmol 2002;133(6):

764–72.

[21] Polito A, Shah SM, Haller JA, et al. Comparison

between retinal thickness analyzer and optical coher-

ence tomography for assessment of foveal thickness

in eyes with macular disease. Am J Ophthalmol 2002;

134(2):240–51.

[22] Master B, Kino G. Confocal microscopy of the eye.

In: Masters BR, editor. Noninvasive diagnostic tech-

niques in ophthalmology. New York: Springer-Ver-

lag; 1990.

[23] Shuman H, Murray J, DiLullo C. Confocal micros-

copy: an overview. Biotechniques 1989;7:154–63.

[24] Kruse FE, Burk RO, Volcker HE, et al. Reproducibil-

ity of topographic measurements of the optic nerve

head with laser tomographic scanning. Ophthalmol-

ogy 1989;96(9):1320–4.

[25] Dreher AW, Tso PC, Weinreb RN. Reproducibility of

topographic measurements of the normal and glau-

comatous optic nerve head with the laser tomographic

scanner. Am J Ophthalmol 1991;111(2):221–9.

[26] Rohrschneider K, Burk RO, Kruse FE, Volcker HE.

Reproducibility of the optic nerve head topography

with a new laser tomographic scanning device. Oph-

thalmology 1994;101(6):1044–9.

[27] Rohrschneider K, Burk RO, Volcker HE. Reproduc-

ibility of topometric data acquisition in normal and

glaucomatous optic nerve heads with the laser tomo-

graphic scanner. Graefes Arch Clin Exp Ophthalmol

1993;231(8):457–64.

[28] Chauhan BC, LeBlanc RP, McCormick TA, Rogers

JB. Test-retest variability of topographic measure-

ments with confocal scanning laser tomography in

patients with glaucoma and control subjects. Am J

Ophthalmol 1994;118(1):9–15.

[29] Janknecht P, Funk J. Optic nerve head analyser

and Heidelberg retina tomograph: accuracy and re-

producibility of topographic measurements in a

model eye and in volunteers. Br J Ophthalmol 1994;

78(10):760–8.

[30] Weinreb RN, Lusky M, Bartsch DU, Morsman D.

Effect of repetitive imaging on topographic measure-

ments of the optic nerve head. Arch Ophthalmol

1993;111(5):636–8.


[31] Garway-Heath DF, Poinoosawmy D, Wollstein G,

et al. Inter- and intraobserver variation in the analysis

of optic disc images: comparison of the Heidelberg

retina tomograph and computer assisted planimetry.

Br J Ophthalmol 1999;83(6):664–9.

[32] Sihota R, Gulati V, Agarwal HC, et al. Variables

affecting test-retest variability of Heidelberg Retina

Tomograph II stereometric parameters. J Glaucoma

2002;11(4):321–8.

[33] Mikelberg F, Wijsman K, Schulzer M. Reproducibil-

ity of topographic parameters obtained with Heidel-

berg Retina Tomograph. J Glaucoma 1993;2:101–3.

[34] Zangwill LM, Berry CC, Weinreb RN. Optic disc

topographic measurements after pupil dilation. Oph-

thalmology 1999;106(9):1751–5.

[35] Hatch WV, Flanagan JG, Williams-Lyn DE, et al.

Interobserver agreement of Heidelberg retina tomo-

graph parameters. J Glaucoma 1999;8(4):232–7.

[36] Iester M, De Ferrari R, Zanini M. Topographic analy-

sis to discriminate glaucomatous from normal optic

nerve heads with a confocal scanning laser: new optic

disk analysis without any observer input. Surv Oph-

thalmol 1999;44:S33–40.

[37] Tan JC, Garway-Heath DF, Fitzke FW, Hitchings

RA. Reasons for rim area variability in scanning laser

tomography. Invest Ophthalmol Vis Sci 2003;44(3):

1126–31.

[38] Burk RO, Vihanninjoki K, Bartke T, et al. Devel-

opment of the standard reference plane for the

Heidelberg retina tomograph. Graefes Arch Clin Exp

Ophthalmol 2000;238(5):375–84.

[39] Park KH, Caprioli J. Development of a novel refer-

ence plane for the Heidelberg retina tomograph with

optical coherence tomography measurements. J Glau-

coma 2002;11(5):385–91.

[40] Jonas JB, Mardin CY, Grundler AE. Comparison

of measurements of neuroretinal rim area between

confocal laser scanning tomography and planime-

try of photographs. Br J Ophthalmol 1998;82(4):

362–6.

[41] Dichtl A, Jonas JB, Mardin CY. Comparison between

tomographic scanning evaluation and photographic

measurement of the neuroretinal rim. Am J Ophthal-

mol 1996;121(5):494–501.

[42] Chauhan BC, McCormick TA. Effect of the cardiac

cycle on topographic measurements using confocal

scanning laser tomography. Graefes Arch Clin Exp

Ophthalmol 1995;233:568–72.

[43] Brigatti L, Caprioli J. Correlation of visual field with

scanning confocal laser optic disc measurements in

glaucoma. Arch Ophthalmol 1995;113(9):1191–4.

[44] Iester M, Mikelberg FS, Courtright P, Drance SM.

Correlation between the visual field indices and Hei-

delberg retina tomograph parameters. J Glaucoma

1997;6(2):78–82.

[45] Eid TM, Spaeth GL, Katz LJ, et al. Quantitative esti-

mation of retinal nerve fiber layer height in glaucoma

and the relationship with optic nerve head topography

and visual field. J Glaucoma 1997;6(4):221–30.

[46] Hatch WV, Flanagan JG, Etchells EE, et al. Laser

scanning tomography of the optic nerve head in ocu-

lar hypertension and glaucoma. Br J Ophthalmol

1997;81(10):871–6.

[47] Iester M, Broadway DC, Mikelberg FS, Drance SM.

A comparison of healthy, ocular hypertensive, and

glaucomatous optic disc topographic parameters.

J Glaucoma 1997;6(6):363–70.

[48] Iester M, Mikelberg FS, Drance SM. The effect

of optic disc size on diagnostic precision with the

Heidelberg retina tomograph. Ophthalmology 1997;

104(3):545–8.

[49] Bathija R, Zangwill L, Berry CC, et al. Detection of

early glaucomatous structural damage with confocal

scanning laser tomography. J Glaucoma 1998;7(2):

121–7.

[50] Mikelberg FS, Parfitt CM, Swindale NV. Ability of

the Heidelberg Retina Tomograph to detect early

glaucomatous visual field loss. J Glaucoma 1996;4:

242–7.

[51] Wollstein G, Garway-Heath DF, Hitchings RA. Iden-

tification of early glaucoma cases with the scanning

laser ophthalmoscope. Ophthalmology 1998;105(8):

1557–63.

[52] Miglior S, Guareschi M, Albe E, et al. Detection of

glaucomatous visual field changes using the Moor-

fields regression analysis of the Heidelberg retina

tomograph. Am J Ophthalmol 2003;136(1):26–33.

[53] Ford BA, Artes PH, McCormick TA, et al. Compari-

son of data analysis tools for detection of glaucoma

with the Heidelberg Retina Tomograph. Ophthalmol-

ogy 2003;110(6):1145–50.

[54] Bowd C, Chan K, Zangwill LM, et al. Comparing

neural networks and linear discriminant functions

for glaucoma detection using confocal scanning laser

ophthalmoscopy of the optic disc. Invest Ophthalmol

Vis Sci 2002;43(11):3444–54.

[55] Kamal DS, Viswanathan AC, Garway-Heath DF,

et al. Detection of optic disc change with the Heidel-

berg retina tomograph before confirmed visual field

change in ocular hypertensives converting to early

glaucoma. Br J Ophthalmol 1999;83(3):290–4.

[56] Chauhan BC, Blanchard JW, Hamilton DC, LeBlanc

RP. Technique for detecting serial topographic

changes in the optic disc and peripapillary retina using

scanning laser tomography. Invest Ophthalmol Vis

Sci 2000;41(3):775–82.

[57] Chauhan BC, McCormick TA, Nicolela MT, LeBlanc

RP. Optic disc and visual field changes in a prospec-

tive longitudinal study of patients with glaucoma:

comparison of scanning laser tomography with con-

ventional perimetry and optic disc photography. Arch

Ophthalmol 2001;119(10):1492–9.

[58] Weinreb RN, Shakiba S, Zangwill L. Scanning laser

polarimetry to measure the nerve fiber layer of nor-

mal and glaucomatous eyes. Am J Ophthalmol 1995;

119(5):627–36.

[59] Hoh S, Greenfield D, Liebmann J, et al. Effect of

pupillary dilation on retinal nerve fiber layer thick-


ness measured by scanning laser polarimetry in

eyes with and without cataract. J Glaucoma 1999;8:

159–63.

[60] Hollo G, Suveges I, Nagymihaly A, Vargha P. Scan-

ning laser polarimetry of the retinal nerve fibre layer

in primary open angle and capsular glaucoma. Br J

Ophthalmol 1997;81(10):857–61.

[61] Weinreb R, Dreher AW, Coleman A, et al. Histo-

pathologic validation of Fourier-ellipsometry mea-

surements of retinal nerve fiber layer thickness.


[62] Greenfield D, Knighton RW, Huang X. Effect of

corneal polarization axis on assessment of retinal

nerve fiber layer thickness by scanning laser polar-

imetry. Am J Ophthalmol 2000;129:715–22.

[63] Knighton RW, Huang X. Linear birefringence of the

central human cornea. Invest Ophthalmol Vis Sci

2002;43:82–6.

[64] Hoh ST, Ishikawa H, Greenfield DS. Peripapillary

nerve fiber layer thickness measurement reproducibil-

ity using scanning laser polarimetry. J Glaucoma

1998;7:12–5.

[65] Zangwill L, Berry CA, Garden VS, Weinreb RN.

Reproducibility of retardation measurements with

the nerve fiber analyzer II. J Glaucoma 1997;6:

384–9.

[66] Waldock A, Potts MJ, Sparrow JM, Karwatowski

WS. Clinical evaluation of scanning laser polarime-

try. I. Intraoperator reproducibility and design of a

blood vessel removal algorithm. Br J Ophthalmol

1998;82(3):252–9.

[67] Poinoosawmy D, Fontana L, Wu J, et al. Variation of

nerve fibre layer thickness measurements with age

and ethnicity by scanning laser polarimetry. Br J Oph-

thalmol 1997;81:350–4.

[68] Tjon-Fo-sang M, Lemij H. Retinal nerve fiber layer

measurements in normal black subjects as determined

with scanning laser polarimetry. Ophthalmology

1998;105:78–81.

[69] Pons ME, Rothman RF, Ozden RG, et al. Vitreous

opacities affect scanning laser polarimetry measure-

ments. Am J Ophthalmol 2001;131(4):511–3.

[70] Colen TP, Lemij HG. Motion artifacts in scann-

ing laser polarimetry. Ophthalmology 2002;109(8):

1568–72.

[71] Colen TP, Lemij HG. Prevalence of split nerve fi-

ber layer bundles in healthy eyes imaged with scan-

ning laser polarimetry. Ophthalmology 2001;108(1):

151–6.

[72] Bagga H, Greenfield DS, Knighton RW. Scanning

laser polarimetry with variable corneal compensation:

identification and correction for corneal birefringence

in eyes with macular disease. Invest Ophthalmol Vis

Sci 2003;44(5):1969–76.

[73] Xu L, Chen PP, Chen YY, et al. Quantitative nerve

fiber layer measurement using scanning laser polar-

imetry and modulation parameters in the detection of

glaucoma. J Glaucoma 1998;7(4):270–7.

[74] Horn FK, Jonas JB, Martus P, et al. Polarimetric

measurement of retinal nerve fiber layer thickness in

glaucoma diagnosis. J Glaucoma 1999;8(6):353–62.

[75] Kogure S, Iijima H, Tsukahara S. A new parameter

for assessing the thickness of the retinal nerve fiber

layer for glaucoma diagnosis. Eur J Ophthalmol 1999;

9(2):93–8.

[76] Essock EA, Sinai MJ, Fechtner RD, et al. Fourier

analysis of nerve fiber layer measurements from scan-

ning laser polarimetry in glaucoma: emphasizing

shape characteristics of the ‘double-hump’ pattern.

J Glaucoma 2000;9(6):444–52.

[77] Hoh ST, Greenfield DS, Mistlberger A, et al. Optical

coherence tomography and scanning laser polarime-

try in normal, ocular hypertensive, and glaucomatous

eyes. Am J Ophthalmol 2000;129(2):129–35.

[78] Bowd C, Zangwill LM, Berry CC, et al. Detecting

early glaucoma by assessment of retinal nerve fiber

layer thickness and visual function. Invest Ophthal-

mol Vis Sci 2001;42(9):1993–2003.

[79] Sinai MJ, Essock EA, Fechtner RD, Srinivasan N.

Diffuse and localized nerve fiber layer loss measured

with a scanning laser polarimeter: sensitivity and spe-

cificity of detecting glaucoma. J Glaucoma 2000;

9(2):154–62.

[80] Choplin NT, Lundy DC. The sensitivity and specific-

ity of scanning laser polarimetry in the detection of

glaucoma in a clinical setting. Ophthalmology 2001;

108(5):899–904.

[81] Poinoosawmy D, Tan JC, Bunce C, Hitchings RA.

The ability of the GDx nerve fibre analyser neural

network to diagnose glaucoma. Graefes Arch Clin

Exp Ophthalmol 2001;239(2):122–7.

[82] Sanchez-Galeana C, Bowd C, Blumenthal EZ, et al.

Using optical imaging summary data to detect glau-

coma. Ophthalmology 2001;108(10):1812–8.

[83] Zangwill LM, Bowd C, Berry CC, et al. Discriminat-

ing between normal and glaucomatous eyes using

the Heidelberg Retina Tomograph, GDx Nerve Fiber

Analyzer, and Optical Coherence Tomograph. Arch

Ophthalmol 2001;119(7):985–93.

[84] Bozkurt B, Irkec M, Karaagaoglu E, Orhan M. Scan-

ning laser polarimetric analysis of retinal nerve fiber

layer thickness in Turkish patients with glaucoma and

ocular hypertension. Eur J Ophthalmol 2002;12(5):

406–12.

[85] Funaki S, Shirakashi M, Yaoeda K, et al. Specificity

and sensitivity of glaucoma detection in the Japanese

population using scanning laser polarimetry. Br J

Ophthalmol 2002;86(1):70–4.

[86] Greaney MJ, Hoffman DC, Garway-Heath DF, et al.

Comparison of optic nerve imaging methods to dis-

tinguish normal eyes from those with glaucoma. In-


[87] Weinreb RN, Bowd C, Zangwill LM. Glaucoma de-

tection using scanning laser polarimetry with variable

corneal polarization compensation. Arch Ophthalmol

2003;121(2):218–24.

[88] Bowd C, Zangwill LM, Weinreb RN. Association

between scanning laser polarimetry measurements


using variable corneal polarization compensation and

visual field sensitivity in glaucomatous eyes. Arch

Ophthalmol 2003;121(7):961–6.

[89] Bagga H, Greenfield DS, Feuer W, Knighton RW.

Scanning laser polarimetry with variable corneal com-

pensation and optical coherence tomography in nor-

mal and glaucomatous eyes. Am J Ophthalmol 2003;

135(4):521–9.

[90] Reus NJ, Colen TP, Lemij HG. Visualization of lo-

calized retinal nerve fiber layer defects with the GDx

with individualized and with fixed compensation of

anterior segment birefringence. Ophthalmology 2003;

110(8):1512–6.

[91] Colen TP, Lemij HG. Sensitivity and specificity of the

GDx: clinical judgment of standard printouts versus

the number. J Glaucoma 2003;12(2):129–33.

[92] Nguyen NX, Horn FK, Hayler J, et al. Retinal nerve

fiber layer measurements using laser scanning po-

larimetry in different stages of glaucomatous optic

nerve damage. Graefes Arch Clin Exp Ophthalmol

2002;240(8):608–14.

[93] Lauande-Pimentel R, Carvalho RA, Oliveira HC,

et al. Discrimination between normal and glaucoma-

tous eyes with visual field and scanning laser polar-

imetry measurements. Br J Ophthalmol 2001;85(5):

586–91.

[94] Medeiros FA, Zangwill LM, Bowd C, et al. Fourier

analysis of scanning laser polarimetry measurements

with variable corneal compensation in glaucoma. In-


[95] Huang D, Swanson EA, Lin CP, et al. Optical coher-

ence tomography. Science 1991;254(5035):1178–81.

[96] Schuman JS, Pedut-Kloizman T, Hertzmark E, et al.

Reproducibility of nerve fiber layer thickness mea-

surements using optical coherence tomography. Oph-

thalmology 1996;103(11):1889–98.

[97] Drexler W, Morgner U, Ghanta RK, et al. Ultrahigh-

resolution ophthalmic optical coherence tomography.

Nat Med 2001;7(4):502–7.

[98] Blumenthal EZ, Williams JM, Weinreb RN, et al.

Reproducibility of nerve fiber layer thickness mea-

surements by use of optical coherence tomography.

Ophthalmology 2000;107(12):2278–82.

[99] Koozekanani D, Roberts C, Katz SE, Herderick EE.

Intersession repeatability of macular thickness mea-

surements with the Humphrey 2000 OCT. Invest Oph-

thalmol Vis Sci 2000;41(6):1486–91.

[100] Konno S, Akiba J, Yoshida A. Retinal thickness

measurements with optical coherence tomography

and the scanning retinal thickness analyzer. Retina

2001;21(1):57–61.

[101] Massin P, Vicaut E, Haouchine B, et al. Reproduc-

ibility of retinal mapping using optical coherence to-

mography. Arch Ophthalmol 2001;119(8):1135–42.

[102] Muscat S, Parks S, Kemp E, Keating D. Repeatability

and reproducibility of macular thickness measure-

ments with the Humphrey OCT system. Invest Oph-

thalmol Vis Sci 2002;43(2):490–5.

[103] Jones AL, Sheen NJ, North RV, Morgan JE. The

Humphrey optical coherence tomography scanner:

quantitative analysis and reproducibility study of

the normal human retinal nerve fibre layer. Br J Oph-

thalmol 2001;85(6):673–7.

[104] Carpineto P, Ciancaglini M, Zuppardi E, et al.

Reliability of nerve fiber layer thickness measure-

ments using optical coherence tomography in normal

and glaucomatous eyes. Ophthalmology 2003;110(1):

190–5.

[105] Gurses-Ozden R, Ishikawa H, Hoh ST, et al. Increas-

ing sampling density improves reproducibility of

optical coherence tomography measurements. J Glau-

coma 1999;8(4):238–41.

[106] Schuman JS, Hee MR, Puliafito CA, et al. Quantifi-

cation of nerve fiber layer thickness in normal and

glaucomatous eyes using optical coherence tomogra-

phy. Arch Ophthalmol 1995;113(5):586–96.

[107] Mok KH, Lee VW, So KF. Retinal nerve fiber layer

measurement of the Hong Kong Chinese population

by optical coherence tomography. J Glaucoma 2002;

11(6):481–3.

[108] Varma R, Bazzaz S, Lai M. Optical tomography-

measured retinal nerve fiber layer thickness in nor-

mal Latinos. Invest Ophthalmol Vis Sci 2003;44(8):

3369–73.

[109] Guedes V, Schuman JS, Hertzmark E, et al. Optical

coherence tomography measurement of macular

and nerve fiber layer thickness in normal and glau-

comatous human eyes. Ophthalmology 2003;110(1):

177–89.

[110] Bowd C, Weinreb RN, Williams JM, Zangwill LM.

The retinal nerve fiber layer thickness in ocular hy-

pertensive, normal, and glaucomatous eyes with op-

tical coherence tomography. Arch Ophthalmol 2000;

118(1):22–6.

[111] Soliman MA, Van Den Berg TJ, Ismaeil AA, et al.

Retinal nerve fiber layer analysis: relationship be-

tween optical coherence tomography and red-free

photography. Am J Ophthalmol 2002;133(2):187–95.

[112] Greenfield DS, Bagga H, Knighton RW. Macular

thickness changes in glaucomatous optic neuropathy

detected using optical coherence tomography. Arch

Ophthalmol 2003;121(1):41–6.

[113] Lederer DE, Schuman JS, Hertzmark E, et al. Analy-

sis of macular volume in normal and glaucomatous

eyes using optical coherence tomography. Am J Oph-

thalmol 2003;135(6):838–43.

[114] Schuman JS, Wollstein G, Farra T, et al. Comparison

of optic nerve head measurements obtained by optical

coherence tomography and confocal scanning la-

ser ophthalmoscopy. Am J Ophthalmol 2003;135(4):

504–12.

[115] Essock EA, Sinai MJ, Bowd C, et al. Fourier analysis

of optical coherence tomography and scanning laser

polarimetry retinal nerve fiber layer measurements in

the diagnosis of glaucoma. Arch Ophthalmol 2003;

121(9):1238–45.

[116] Zeimer RC, Mori MT, Khoobehi B. Feasibility test

of a new method to measure retinal thickness non-


invasively. Invest Ophthalmol Vis Sci 1989;30(10):

2099–105.

[117] Zeimer R, Shahidi M, Mori M, et al. A new method

for rapid mapping of the retinal thickness at the pos-

terior pole. Invest Ophthalmol Vis Sci 1996;37(10):

1994–2001.

[118] Neubauer AS, Priglinger S, Ullrich S, et al. Compari-

son of foveal thickness measured with the retinal

thickness analyzer and optical coherence tomography.

Retina 2001;21(6):596–601.

[119] Fritsche P, van der Heijde R, Suttorp-Schulten MS,

Polak BC. Retinal thickness analysis (RTA): an

objective method to assess and quantify the retinal

thickness in healthy controls and in diabetics without

diabetic retinopathy. Retina 2002;22(6):768–71.

[120] Polito A, Shah SM, Haller JA, et al. Comparison

between retinal thickness analyzer and optical co-

herence tomography for assessment of foveal thick-

ness in eyes with macular disease. Am J Ophthalmol

2002;134(2):240–51.

[121] Zeimer R, Asrani S, Zou S, et al. Quantitative detec-

tion of glaucomatous damage at the posterior pole by

retinal thickness mapping: a pilot study. Ophthalmol-

ogy 1998;105(2):224–31.

[122] Mistlberger A, Liebmann JM, Greenfield DS, et al.

Heidelberg retina tomography and optical coherence

tomography in normal, ocular-hypertensive, and

glaucomatous eyes. Ophthalmology 1999;106(10):

2027–32.


High-resolution functional optical imaging: from the

neocortex to the eye

Amiram Grinvald, PhDa,*, Tobias Bonhoeffer, PhDb, Ivo Vanzetta, PhDc,Ayala Pollack, MDd, Eyal Aloni, MDd, Ron Ofri, MDe, Darin Nelson, PhDc

aDepartment of Neurobiology, The Weizmann Institute of Science, The Grodetsky Center for Research of Higher Brain Functions,

Rehovot, 76100, IsraelbDepartment of Neurobiology, The Max Plank Institute, Am Klopferspitz 18, 82152 Muenchen-Martinsried, Germany

cOptical Imaging, 2 Bergman Street, Rehovot, 76705, IsraeldDepartment of Ophthalmology, Kaplan Hospital, Rehovot, 76100, Israel

eKoret School of Veterinary Medicine, The Hebrew University of Jerusalem, PO Box 12, Rehovot, 76100, Israel

Progress in the fight against blindness depends on ischemia. No currently existing technology can di-

the development of new technologies enabling early

detection and accurate monitoring of diseases such as

diabetic retinopathy, age-related macular degenera-

tion, and glaucoma. New technology may also con-

tribute to better understanding of the pathogenesis of

these diseases, leading to the development of new

drugs and the improvement of preventive treatment.

Extensive research has led to the evolution of several

technologies in the last few decades [1–18]. Never-

theless, the technologies currently in use do not offer

comprehensive information. Direct blood flow visu-

alization and quantification remain elusive. To date,

no technology has been able to image retinal blood

flow directly, particularly in the capillaries. Because

many diseases affect small vessels, information re-

garding blood flow, particularly in the capillaries, is

needed for early diagnosis of retinal diseases. Oxime-

try measuring the intravascular oxygen content is an

essential parameter when assessing the degree of

tissue vitality. Many retinal diseases involve blood

vessels, resulting in a reduction of the flow of oxygen

and nutrients to the retinal tissue and leading to


doi:10.1016/j.ohc.2003.12.003

Darin Nelson and Ivo Vanzetta are employees of Optical

Imaging. Amiram Grinvald has financial interest in Optical

Imaging Ltd.

* Corresponding author. Department of Neurobiology,

The Weizmann Institute of Science, Rehovot, 76100, Israel.

E-mail address: Amiram.Grinvald@weizmann.

weizmann.ac.il (A. Grinvald).

rectly image oximetric parameters in the retina [1] in

a quantitative rapid manner [19]. There are no means

for the direct assessment of oxygen, the key pa-

rameter of tissue viability. No technology has been

able to demonstrate directly the metabolic functional

status of components of the retinal tissue. Once dis-

covered, functional or metabolic signals will provide

necessary information on the viability and vitality of

diseased tissues using direct methods rather than

indirect methods.

Functional optical imaging, a noninvasive techno-

logy widely used in brain research, has recently been

adapted for functional retinal imaging. The authors

believe that functional parameters are more promising

for providing the earliest possible diagnostics rather

than anatomic parameters. This article briefly reviews

functional optical imaging in the neocortex and

describes the type of results obtained by applying

this technology to functional imaging of the eye.

Preliminary results have been published in abstract

form [20–24].

Neocortical functional optical imaging

Noninvasive functional optical imaging of the neo-

cortex based on intrinsic signals without the use of a

contrast agent was developed more than a decade ago

[25–27]. This technology has been designed to detect

minute optical differences between cortical images

s reserved.

A. Grinvald et al / Ophthalmol Clin N Am 17 (2004) 53–6754

obtained under different neuronal processing condi-

tions, differences that previously were thought to be

invisible. Unprecedented accurate functional mapping

of the functional architecture of the neocortex can be

performed. Using high-resolution digital imaging of

exposed cortex and removing extensive biologic

noise, it has been possible to create information-rich

images that reveal minute changes in the light intensity

reflected from the cortex, changes indicating intrinsic

metabolic activity of activated cortical domains (eg,

changes in oxygen consumption, blood flow, blood

volume, and light scattering). The imaging setup is

depicted in Fig. 1. Imaging begins with a sequence of

digital images of the target organ, taken with a video or

digital camera, just after the cortex has been activated

with a given sensory input. Subsequently, functional

maps are obtained by analyzing the small differ-

ences between the series of images owing to the

increased oxygen consumption, blood flow, and ac-

tivity-dependent light-scattering signals. This stage is

followed by a biologic system–specific noise reduc-

tion procedure that enables the extraction of small but

relevant signals from large artifacts (some are irrele-

vant physiological events that are removed), image

enhancement, and quantification.

Functional optical imaging based on intrinsic signals

Neocortical functional optical imaging based on

intrinsic signals has provided a new level of under-

standing of the principles underlying cortical devel-

opment, organization, and function [28]. In the best

cases, it provides a spatial resolution of 20 mm for

mapping the layout of cortical columns in vivo. This

excellent resolution is much better than that required

to resolve cortical columns. The discussion herein

provides a brief overview of the types of applications

that have been pursued, and the general implications

of some findings for other neuroimaging techniques

based on hemodynamic responses, particularly func-

tional magnetic resonance imaging (f-MRI).

In the mammalian brain, cells that perform a given

function or that share common functional properties

are often grouped together in cortical columns running

from pia to white mater [29–31]. Examples include

the orientation and ocular dominance columns of the

primary visual cortex. Without knowing the function

performed by each group of neurons, it is difficult to

discover the principles underlying the neural code and

its implementation. Attaining an understanding of the

functional organization of a given cortical area is a key

step toward revealing the fundamental mechanisms of

information processing therein. Of special importance

are experimental methods that allow visualization of

the functional organization of cortical columns, par-

ticularly methods providing high spatial and temporal

resolution. To date, single- or multi-unit extracellular

recording techniques have provided the best tools for

studying the functional response properties of single

cortical neurons and their synaptic interactions. Al-

though multi-electrode techniques are more promis-

ing, the size and placement of these electrode arrays

pose severe problems. Several imaging techniques

have been developed that yield information about

the spatial distribution of active neurons in the brain.

Each of these techniques has significant advantages as

well as limitations. Although the 2-deoxyglucose

method permits postmortem visualization of active

brain areas, or even of single cells, it is a one shot

approach. Only a single stimulus condition in a single

animal can be assayed (the two-isotope method per-

mits the mapping of activity resulting from two

stimulus conditions). Positron-emission tomography

and f-MRI offer spectacular three-dimensional local-

ization of active regions in the functioning human

brain, but both methods have low spatial resolution,

two to three orders of magnitude worse than optical

imaging. Other imaging techniques have also been

applied in vivo with success but suffer from a limited

spatial resolution, a limited temporal resolution, or

both. Among these methods are radioactive imaging

of changes in blood flow, electroencephalography,

magnetoencephalography, and thermal imaging.

Overview of functional optical imaging in the

neocortex

Currently, the best method for imaging the func-

tional architecture of the cortex is based on slow

changes in the intrinsic optical properties of active

brain tissue. This method permits visualization of

active cortical regions at a spatial resolution better

than 50 mm and is immune to some of the problems

associated with the use of optical imaging based on

extrinsic probes [32–35]. The sources for these ac-

tivity-dependent intrinsic signals include changes in

the physical properties of the tissue affecting light

scattering, and changes in the absorption, fluores-

cence, or other optical properties of intrinsic mole-

cules. The fact that, in many tissues, small intrinsic

optical changes are associated with metabolic activity

has been known since the pioneering experiments of

Kelin [36] and Millikan [37] on the absorption of

cytochromes and hemoglobin, respectively. The first

optical recording of neuronal activity was made al-

most 50 years ago by Hill and Keynes [38], who

detected light-scattering changes in active nerves.

Changes in the absorption or fluorescence of intrin-

Fig. 1. Optical imaging of functional maps in vivo. (A) The setup. Digital CCD images are taken of the exposed cortex of the

animal, which is sealed in an oil-filled chamber. The cortex is illuminated with light of 605 nm wavelength. During image

acquisition, the animal is visually stimulated with moving gratings projected onto a frosted glass screen by a video projector. The

acquired images are digitized in a camera controller, which transfers the data to the computer controlling the entire experiment.

Functional maps are subsequently analyzed and are displayed on a color monitor. To determine the quality of the maps during the

imaging sessions, the data can be sent to a second computer for detailed, quasi on-line analysis. (Data from Ts’o DY, Frostig RD,

Lieke E, Grinvald A. Functional organization of primate visual cortex revealed by high resolution optical imaging. Science

1990;249:417–20.) (B) Imaging functional architecture. An activity map for one orientation is obtained straightforwardly by

dividing the image captured during presentation of this orientation by the average of the images captured during presentation of

all orientations. The image of the cortical surface illuminated with green light to emphasize the vasculature is shown at the top.

Activity maps evoked by visual stimulation with horizontal and vertical gratings are shown at the bottom. Scale bar: 1mm.

(Data from Bonhoeffer T, Grinvald A. Iso-orientation domains in cat visual cortex are arranged in pinwheel-like patterns. Nature

1991;353:429–31.)

A. Grinvald et al / Ophthalmol Clin N Am 17 (2004) 53–67 55


sic chromophores were extensively investigated by

Chance and colleagues [39] and Jobsis [40] (for a

review, see the article by Cohen [41]).

The basic experimental setup for optical imaging

experiments is shown in Fig. 1. The anesthetized

animal is held in a stereotaxic frame (not shown).

The exposed brain is illuminated with flexible light

guides, and digital pictures are acquired by the cam-

era, which views the exposed cortex through a cranial

window. The data are analyzed on the computer

controlling the experiment, and the results are dis-

played on a color monitor (for detailed practical

guidelines, see the article by Grinvald et al [28]).

Sources of intrinsic signals

More than a century ago, Roy and Sherrington

[42] postulated that ‘‘the brain possesses an intrinsic

mechanism by which its vascular supply can be

varied locally in correspondence with local variations

of functional activity.’’ Contemporary imaging tech-

niques have indeed demonstrated a strong coupling

between neuronal activity and local metabolic activ-

ity and blood flow [25,43,44]. More recently, Ogawa

and colleagues [45] have found that the intrinsic

signal is useful for functional brain mapping with

MR imaging. Optical imaging has revealed the exact

spatiotemporal characteristics of this coupling.

Although the intrinsic signal has different compo-

nents that originate from different sources, the func-

tional maps obtained at different wavelengths are very

similar, and it appears that all of these components can

be used for functional mapping [27]. Malonek and

Grinvald [46] have used optical imaging spectroscopy,

a new technique providing simultaneous spectral

information from many cortical locations in the form

of a spatiospectral image. The images obtained with

imaging spectroscopy show the spectral changes at

many wavelengths for each cortical point (y versus l)along a selected line. Using this technique, the authors

have measured the spatial, temporal, and spectral

characteristics of light reflected from the surface of

the visual cortex following natural stimulation. The

technical details are published elsewhere [46–48].

The sources of the intrinsic signals in reflection

measurements from the neocortex are the result of the

following cellular or tissue changes:

� Activity-dependent oxygen consumption affect-

ing the hemoglobin saturation level� Changes in blood volume affecting tissue light

absorption� Changes in blood volume affecting tissue light

scattering

� Changes in blood flow affecting the hemoglobin

saturation level� Activity-dependent changes in tissue light scat-

tering

The first component of the intrinsic signal in the

neocortex originates from activity-dependent changes

in the oxygen saturation level of hemoglobin. This

change in oxygenation contains two different compo-

nents. The first component is an early event—an

increase in the deoxyhemoglobin concentration result-

ing from elevated oxygen consumption of the neurons

owing to their metabolic activity. This component

causes a darkening of the cortex owing to increased

absorption by deoxyhemoglobin molecules and is

referred to as the initial dip. The second component

originates from an activity-dependent increase in

cortical blood volume, increasing the amount of light

absorbed by activated cortical regions in a wave-

length-dependent fashion following the absorption

spectra of hemoglobin. The level of spatial regulation

of this component has been of great interest, because

most f-MRI studies use this component of the hemo-

dynamic response. In previous studies of optical

imaging and imaging spectroscopy in the visual cortex

of cats and monkeys, the authors reported a much

weaker colocalization of cortical blood volume

changes with the electrically active columns than the

[Hbr] increase during the initial dip [27,46–49]. This

poorer colocalization is revealed by a lower signal-to-

noise ratio in spatial maps of activation calculated

from the cortical blood volume component when

compared with those from the initial dip. This obser-

vation is especially true for maps in which the re-

sponse to a single stimulus condition is normalized by

the response to a blank (single-condition maps). For

differential imaging, in which the response to a

stimulus condition is normalized by or subtracted

from the response to an orthogonal stimulus condi-

tion, the maps from the cortical blood volume are still

somewhat poorer than from the initial dip; however,

high-quality functional cortical blood volume maps

have been obtained in cat and monkey visual cortex

with differential imaging [27]. Such differential pro-

cedure removes much of the cortical blood volume

noise, because it nearly eliminates the contribution of

the nonstimulus-specific part of the vascular response.

These results indicate that part of the cortical blood

volume changes colocalize with areas of increased

neuronal activity.

The third component is a delayed event—an ac-

tivity-related increase in blood flow causing a de-

crease in the deoxyhemoglobin concentration. The

blood rushing into the activated tissue contains higher


levels of oxyhemoglobin. Another signal component

originates from changes in blood volume, probably

owing to local capillary recruitment or a rapid filling

of capillaries and dilation of venules in an area con-

taining electrically active neurons. These blood-related

components dominate the signal at wavelengths

between 400 and 630 nm. It has recently been shown

that capillaries blood volume changes co-localize with

electrical activity and can be used to obtain functional

maps [71]. However, blood volumes in larger vessels

do not. Blood volume changes from all the microvas-

cular compartments compress the cortical tissue, giv-

ing rise to a change in light scattering that exists at all

wavelengths, and exhibit a time course typical of

blood volume changes rather than light-scattering

changes that originate from electrical activity per se.

The last significant component of the fast and slow

intrinsic signal arises from changes in light scattering

that accompany cortical activation [32,50,51]. These

changes are caused by ion and water movement,

expansion and contraction of extracellular spaces,

capillary expansion, or neurotransmitter release.

Light-scattering components become a significant

source of intrinsic signals above 630 nm and prob-

ably dominate the intrinsic signals in the near-

infrared region above 800 nm.

Studies of the cortical functional architecture: proof

of concept

Initial optical imaging studies investigated well-

known structural elements of the functional architec-

Fig. 2. Functional optical imaging has been applied to multiple co

sensory modalities. Such maps are depicted on the cover of severa

ture, such as ocular dominance in the primary visual

cortex and the ‘‘stripes’’ in V2 [26] or the pinwheel-

like organization of orientation preference [52].

Subsequently, methodologic improvements made it

possible to investigate more subtle features of cortical

organization, such as direction-selective columns or

spatial frequency columns [53–56]. Similar progress

has been obtained in exploring other visual areas. It

has even been possible to demonstrate functional col-

umns in visual areas further up the processing stream,

that is, in areas V4 [57] and MT [53]. Tanaka and

colleagues [58] have used this method to image the

functional organization in the inferotemporal area,

one of the final stages of the visual pathway critical

for object recognition. Extraordinary high-resolution

maps were reported in the tree shrew [59] and ferret

[60]. Although, to date, most optical imaging studies

have been performed in the visual cortex, this is

not the only sensory system that can be studied using

this method. Indeed, functional optical imaging has

proved to be a useful tool for investigating functional

architecture in the somatosensory cortex of the rat

[25,61] and monkey [56] and the auditory cortex of

the guinea pig and gerbil [62–64]. Extraordinary

high-resolution maps have been reported in the ol-

factory bulb [65]. Functional optical imaging has

also been applied to the human brain noninvasively

[66–68]. Fig. 2 depicts a collection of journal and

book covers illustrating how popular this technique

has become in cortical research.

Optical imaging has resolved an outstanding

question that could not have been answered using

rtical and noncortical areas in multiple species and multiple

l journals.

Fig. 3. Relationships among pinwheels, ocular dominance columns, and blobs in primary visual cortex of the macaque monkey.

(A) The optical map of ocular dominance. The dark bands represent columns dominated by input from the right eye. Scale bar:

500 mm. (B) The borders of the ocular dominance columns (black lines taken from A) were overlaid onto the discrete ‘‘angle

map’’ for orientation preference. The pinwheel centers were marked with circles. (C) The continuous angle map of orientation

preference. Same unsmoothed digital data shown in B are depicted using the color scale shown above the map. The oriented

color bars at the right indicate the color code. (D) The CO-rich blobs were marked on the histologic photograph. (E) Angle map

from B and two overlays showing the relationships among blobs, iso-orientation domains, and ocular dominance columns.

( F) Two adjacent fundamental modules, magnified from E. (G) The revised ice cube model illustrating the relationships found in

the above maps. Black lines mark the borders between columns of neurons that receive signals from different eyes. White ovals

represent groups of neurons responsible for color perception (blobs). The pinwheels are formed by neurons involved in the

perception of shape, with each color marking a column of neurons responding selectively to a particular orientation in space.

Note that both the blobs and the centers of the pinwheels lie at the center of the R or L columns. The iso-orientation lines

(appearing as a border between two colors) tend to cross borders of ocular dominance columns (black lines) at right angles. The

slice above the ice cube model depicts two adjacent fundamental modules (400 mm � 800 mm). Each module contains a complete

set of about 60,000 neurons, processing all three features of orientation, depth, and color. (Data from Bartfeld E, Grinvald A.

Relationships between orientation preference pinwheels, cytochrome oxidase blobs and ocular dominance columns in primate

striate cortex. Proc Natl Acad Sci USA 1992;89:11905–9.)



alternative approaches. After Hubel and Wiesel [31]

and Livingstone and Hubel [69] described three sub-

systems responsible for the perception of shape,

depth, and color that combine in the macaque primary

visual cortex in what has been referred to as the ‘‘ice-

cube’’ model, many other groups attempted to ex-

plore the geometrical relationships among these three

subsystems, if any. Optical imaging revealed how the

various columnar subsystems were organized with

respect to one another. Fig. 3 shows the experimental

data and revised ice-cube model, depicting in a

schematic fashion the relationships among orientation

columns, ocular dominance columns, and the blobs in

monkey primary visual cortex that have been found

by the authors [70]. The following detailed relation-

ships have been found:

1. Orientation preference is organized radially.

2. Orientation domains are continuous and have

fuzzy boundaries.

3. Iso-orientation lines tend to cross ocular domi-

nance borders at 90 degrees.

4. Orientation pinwheels are centered on ocular

dominance columns.

5. Blobs are centered on ocular dominance

columns.

6. The centers of the blobs and the centers of the

orientation pinwheels are segregated.

7. A regular mosaiclike organization exists for

each type of functional domain, without there

being an overall pattern of repeating hyper-

columns. The latter finding is probably related to

the existence of short-range (<1 mm) rather than

long-range interactions during development.

Retinal functional optical imaging

Overview

The retinal functional imager (RFI) (Optical Im-

aging Ltd., Rehovot, Israel), based on similar ap-

proaches used in the brain, is a digital imaging system

designed to capture accurate retinal images via a

standard fundus camera. Four key parameters are

offered by the new imager.

1. Blood flow: The RFI can map blood flow velo-

city in retinal vessels, including minute capil-

laries, often only 5 mm in diameter. The system

can accurately measure the blood flow or speed

at any position in the relevant vessel and can

compare this result with the findings in other

vessels, other eyes, or the same patient at ano-

ther point in time.

2. Microvasculature mapping: The RFI can map

otherwise invisible capillaries and highlight

microvasculature pathology, and may prove to

be effective for earlier detection of neovascula-

rization and for follow-up.

3. Oximetry: The RFI can map the oximetric state

of the eye, an important parameter for diagno-

sis, monitoring, and treatment guidelines.

4. Metabolic function: The RFI can track dys-

function in some elements of the metabolic

machinery of the retina, a parameter that may

lead to improved diagnosis and treatment.

The RFI performs these measurements or any

contrast agent in a noninvasive manner with no need

to inject dye. The RFI can also serve as a digital fundus

camera for standard angiographic techniques requiring

intravenous injection of dye, such as fluorescein and

indocyanine green. Capabilities of the RFI have rele-

vance ranging from the exploration of basic issues in

retinal physiology to clinical issues in diseases such as

diabetic retinopathy, glaucoma, age-related macular

degeneration, and other retinal disorders.

Current technology used for mapping neocortical

function allows unprecedented sensitivity in functional

mapping of the retina with respect to the previously

described parameters. Starting from digital images

of the retina in the form of a short movie, minute

previously undetectable changes in reflected light

intensity are revealed. The series of digital images

captured at 20 to 100 Hz are converted into func-

tional maps for the various parameters by extracting

and amplifying subtle differences between each of

the images. One obvious advantage of this system is

that all of the parameters are derived from measure-

ments with a single system. The versatility of the

RFI enables it to be used as a core piece of equip-

ment, suitable for many different clinical examina-

tions and research projects with human subjects or

animal models.

Technical aspects of the retinal functional imager

The RFI system combines digital fundus imaging

with functional optical imaging, providing a turnkey

system with flexible parameter settings and open

standard data storage.

Hardware modification to a standard fundus camera

To measure the previously described parame-

ters, the following enhancements are used for the

fundus camera:


� Fast stroboscopic illumination of the retina is

performed rather than a single flash at a time.

The flash power supply delivers trains up to

8 flashes at up to 100 Hz. This capability is

useful for any application in which rapid ac-

quisition of similar retinal images is important.� Rapid switching is performed among up to four

illumination wavelengths, acquiring multiple

wavelength images with little or no global

movement of the retina.� The 1:5 digital port zooming range, perform-

ing independently of the fundus camera’s in-

ternal field of view selector, lets the operator

choose the level of detail required for the protocol

in question.� A high-speed, large dynamic range megapixel

digital camera is incorporated.

Imaging specifications

Imaging specifications are as follows:

Pixel resolution, 1024 � 1024

Imaged retinal area, 6–60 degrees

Pixel size on retina, 2–20 mmFrame rate (full), up to 50 Hz

Frame rate (binned), up to 100 Hz

Wavelength range, 450–1100 nm

The RFI permits the operator to define specific

data acquisition protocols for measuring parameters

such as flow or oximetry, for standard protocols such

as fluorescein angiography, or for sessions that com-

bine the RFI’s functionalities in novel paradigms

tailored by an individual investigator.

Illumination specifications

Illumination specifications are as follows:

High-intensity flashes, 8 per train

High-intensity flash rate, up to 100 Hz

Filter positions, 4 standard

Filter switching rate, 55 msec

Continuous illumination, 75 W

Imaging and analysis software

The system software contains several components.

An application that supervises image data acquisition

also includes a live view capacity, useful not only for

adjusting alignment and focus during imaging but also

for live viewing of the retina under infrared illumina-

tion. A browser is integrated, providing immediate

feedback on newly acquired images. It also performs

operations such as image re-registration (automatic or

manual), image filtering, and creating and playing

back flow movies. An animation component inte-

grated in the browser allows viewing the results of a

given imaging session with an interactive frame rate

and color table control. It also applies low-pass,

high-pass, and clipping operations to each image as

it is read.

Visualizing retinal blood flow

Invisible information is hidden in a red-free series

of images of the retinal vasculature, taken with the

stroboscopic illumination. Information on how blood

is flowing through the retina’s veins, arteries, and

capillaries can be displayed as a movie that directly

shows individual clusters of red blood cells in mo-

tion. Fig. 4C shows one flow differential image taken

from the eye of a healthy human subject. The movies

can be viewed in a continuous loop by accessing

http://www.weizmann.grinvald.retinalflow. Fig. 4A

shows a red-free (green illuminated) digital image

of normal human macula. This image is one of a

series of six pictures obtained with the RFI over a

period of 100 ms. The overall retinal area imaged was

4.2 mm on a side (about 15 degrees). The yellow

square shows a large area of interest, 1140 mm on a

side (area shown in Figs. 4B and C). The red square

shows a region of interest 264 mm on a side. Fig. 4C

shows part of the corresponding flow image (yellow

in Fig. 4A), extracted using the RFI’s analysis soft-

ware. Interpretation of this flow image is straightfor-

ward. A black spot represents a blood cell or cluster

of red blood cells. In white spots, or gaps, red blood

cells are absent. The courses of several in-focus

capillaries are clearly traced out by trains of black

and white spots, which are elongated by constriction.

Larger blood vessels also have clear gaps and clus-

ters, but these areas are more punctate. In the spaces

between in-focus vessels, the grainy texture of the

image reflects the distribution of red blood cells

moving through deeper, out-of-focus blood vessels.

Quantifying retinal blood motion

The ability of the RFI to visualize individual reti-

nal red blood cells or red blood cell clusters directly,

without the injection of dye, opens up many new

possibilities in the investigation of retinal blood flow,

particularly in small arteries and veins and in capil-

laries. The flow-enhanced movies of the RFI clearly

reveal the motion of individual clusters of red blood

cells—or even single red blood cells—in vessels as

small as capillaries without the confounding conse-

quences of intermediate assumptions and complicated

transformations. These movies provide a tool for the

http:\\www.weizmann.grinvald.retinalflow

Fig. 5. Direct automatic determination of flow rate map in the cat retina. (A) Flow rate map calculated from 200 frames, such as

the one shown in the right panel (B), were collected at 100 Hz using a continuous stroboscopic illumination. The velocity at each

pixel was calculated directly by the program. See text for more details.

Fig. 4. RFI direct visualization of blood flow in the human retina. (A) Red-free image of the fundus of a normal subject. Yellow

highlight shows the region magnified in panels B and C. (B) Red-free image at higher magnification. (C) Single frame of a flow

movie. Small red square shows region of additional magnification, illustrated in panel D. (D) Top row shows standard and

annotated (bottom row) flow images from the subregion corresponding to the red square in A, B, and C, obtained at 20-ms

intervals. The top row shows extracts from the ‘‘raw’’ flow image. Dark clusters of red blood cells appear at lower left, move

rightward, around a sharp curve, and then upward, and exit the frame at the right top. Some motion in other capillaries can also

be seen. In the bottom row, the same images have been annotated for emphasis. The main capillary has been traced in black, and

five gaps between red blood cell clusters have been assigned five colors to emphasize the motion of blood from one frame to the

next. Clusters (black) and gaps (white) move through the capillary shown here at an average rate of 1.6 mm/s.


Fig. 6. Semi-automatic quantification of flow in human retina. (A) Red-free image of a normal subject. This image shows several

arteries (red set of colors) and veins (violet set) that were selected for quantification. Each shade of color corresponds to a certain

velocity range. (B) Samples of the calculated velocity, as discussed in the text. Scale bar: 1mm.

Fig. 7. Enhanced visualization of hardly visible vessels. (A) Expanded view of the yellowed region shown in Fig. 4B. Some fine

venules and arterioles are visible, but very little of the capillary bed itself. (B) Same region ‘‘flow enhanced.’’ Regions with the

strongest flow signal are black, whereas weaker flow regions are white. The flow through larger vessels, although distinct, is less

regimented than the constricted flow through capillaries and appears in gray. (C,D) Similar results obtained from a larger area.

Scale bar: 1mm.


Fig. 8. Oximetric image obtained with the RFI in the retina

of a 36-year-old healthy subject. Due to the different oxygen

saturation of arterial, venous, and capillary blood, arteries,

veins, and capillaries appear in different gray levels, with

white coding for highest and black for lowest oxygen

saturation. The grayish elongated patches are probably arti-

facts from large chroidal vessels and underscore the current

difficulties with the simple spectral analysis presented here.


qualitative assessment of retinal blood flow rates.

Nevertheless, quantification of the characteristics of

these movies is required to reach the ultimate goal of

creating measures of retinal blood dynamics. Such

measures would be useful for detecting and monitor-

ing clinical diseases, for investigating the effects of

change on physiologic parameters, such as the eleva-

tion of intraocular pressure on hemodynamics, and for

following the effects of drugs on retinal ischemia.

The software analysis suite has built-in tools for

flow rate quantification. It supports three methods of

flow rate quantification, fully manual, fully automatic,

and supervised automatic, each appropriate for differ-

ent circumstances. The program also makes the data

available in an open format, allowing researchers to

develop their own analysis algorithms and techniques.

Manual quantification

Manual measurement of the flow rate of identified

red blood cell clusters, although cumbersome, is the

simplest most assumption-free method of quantifying

the retinal blood flow rate. By drawing a segment

connecting the same cluster or gap as it appears in two

sequential images, the instantaneous velocity between

those two frames can be directly determined. Al-

though this method provides excellent precision in

capillaries, manual quantification in much larger ves-

sels is problematic. Red blood cell clusters are free to

combine, separate, and overlap, making one-to-one

identifications of moving spots among images diffi-

cult. Automated correlation analysis can overcome

these problems.

Automatic quantification

In tapetal animals, retinal reflectance is high

enough to permit retinal flow imaging with incandes-

cent light or low-power flashes (continuous trains for

hours), permitting high-speed imaging of hundreds of

images per series or even live viewing of the blood

flow on a computer monitor. One can apply a fully

automatic quantification algorithm to such large frame

numbers. This algorithm looks for correlations among

image pixels at different time offsets. The points with

the closest correlation are matched, and the flow rate

is calculated from this pairing. Fig. 5 shows an

example from the cat retina.

Supervised quantification

By using the constraining paths given by the image

of the blood vessels, one can greatly improve the

signal-to-noise ratio of the time–offset correlation

matching technique. Automatic measurement of flow

rates can be based on as few as six images. The user

loosely traces over the extent of vessels to be ana-

lyzed. The program then automatically identifies them

and determines the blood flow rate within, even de-

tecting variations in flow along the vessels’ length.

Supervised quantification is interactive, enhancing

confidence in the results. The time from initial record-

ing to flow-rate histogram takes a few minutes. Fig. 6

shows an example of such a result.

Visualization of invisible vessels

Many of the capillaries are traced by trains of

clusters and gaps and are nearly invisible in the cor-

responding region of the original red-free image,

because the spatial resolution of the normal eye is

10 to 15 mm. The RFI can noninvasively document

fine details of vascular anatomy that otherwise may

be invisible, even in the sharpest red-free images.

Figs. 7A and B emphasize this point. Figs. 7C and D

provide another example from a larger retinal area.

Determination of the retinal oximetric state

The difference between the absorption spectra of

oxy- and deoxyhemoglobin can be used to determine

the oxygenation of blood with spectroscopic meth-

ods. For a recent review of previous methodologies,

the reader is referred to the article by Harris et al [19].

When the RFI is used in a multi-wavelength mode,

Fig. 9. Qualitative oximetric retinal maps for detection of ischemia in diabetic patients. Each of the three images includes a

comparison of the color image of the retina, the corresponding fluorescein angiogram, and the oximetric maps obtained by the

RFI. (A) In these three panels, it is evident that areas of leakage seen on the fluorescein angiogram are indeed ischemic. (B) The

darker color of the regions near the fovea suggests that the macula is somewhat anoxic. (C) Another example wherein the red-

free and fluorescein angiogram clear sign for anoxia are not seen in another diabetic patient. Note that in B and C, some of the

abnormalities detected by the RFI are hardly seen or difficult to note in the corresponding angiographic images. These latter

results remain to be validated by direct oxygen tension measurements.


spectroscopic decomposition can be performed, and

the oximetric state of the retina can be qualitatively

assessed. Nevertheless, qualitative mapping of the

oximetric state of the retina is rather limited, and

the ultimate goal should be quantitative oximetry. In

some cases, even qualitative mapping can be of

clinical value, particularly in diabetic retinopathy,

Fig. 10. Time course image series of the functional signal from cat

were evoked by local photic stimulation in the visible wavelength

delivered to the cat’s eye, video images of the feline retina were tak

to left, row-wise from top to bottom. Stimulus duration is mark

stimulation is shown at the left. The time course of the signal and

known hemodynamic response. The origin of this metabolic signa

when the detection of patches of anoxic regions can

guide treatment. Fig. 8 depicts qualitative mapping of

the oximetric state obtained with the RFI in the retina

of a 36-year-old healthy subject. As expected, the

oxygen levels in the three major compartments,

arteries, and veins, and capillaries, are different but

rather uniform. In contrast, Fig. 9 shows the findings

retina. Near-infrared local changes (right) in retinal reflection

range. While a 1-s, local, 100-Hz train of green flashes was

en in the infrared at 400 ms/frame. Time increases from right

ed as a black bar. The control time series without phonic

its wavelength dependency indicate that it is not the well-

l remained to be explored. Each image width is 8mm.


in patients with diabetic retinopathy. Each of the three

images includes a comparison of the color image of

the retina, the corresponding fluorescein angiogram,

and the oximetric map obtained by the RFI. A clear

correlation is seen with the finding on fluorescein

angiograms. In the bottom panel, regions in which the

fluorescein angiograms appear normal are revealed as

partially ischemic, underscoring the significance of

qualitative oximetric imaging.

The retinal functional signal

The RFI has been used to measure a novel retinal

functional signal originating from reflectance changes

in the cat retina, evoked on visual stimulation. Its

spatiotemporal properties suggest that it can be attri-

buted to metabolic rather than hemodynamic process-

es or photoreceptor bleaching. In the cat, this signal

tightly colocalizes with the spatial extent of the visual

stimulus. The functional signal depicted in Fig. 10,

obtained in the cat, illustrates the flexibility and power

of the RFI as a tool for making basic research

discoveries in the retina.

Summary

Much work remains to be done to establish the

clinical usefulness of the RFI for early diagnosis and

treatment guidance. The discoveries obtained by func-

tional optical imaging of the neocortex in the last

15 years and the recent RFI studies of the eyes of

normal subjects and patients with diabetic retinopa-

thy, glaucoma, and age-related macular degeneration

suggest that functional optical imaging of the retina

is likely to become a multi-modality powerful clini-

cal tool.

Acknowledgments

The authors thank their previous coworkers, R.

Frostig, E. Lieke, C.D. Gilbert, T. Wiesel, D. Ts’o, R.

Malach, D. Malonek, and A. Shmuel, for their con-

tributions to the functional mapping of the neocortex,

and M. Belkin for his contribution to research with

the RFI.

References

[1] Delori FC. Noninvasive technique for oximetry of

blood in retinal vessels. Appl Opt 1988;27:1113–25.

[2] Kruse F, Burk R, Volcker H, et al. Reproducibility of

topographic measurements of the optic nerve head with

laser tomographic scanning. Ophthalmology 1989;96:

1320–4.

[3] Huang D, Swanson EA, Lin CP, Schuman JS, Stinson

WG, Chang W, et al. Optical coherence tomography.

Science 1991;254:1178–81.

[4] Hee MR, Izatt JA, Swanson EA, Huang D, Lin CP,

Schuman JS, et al. Micron resolution imaging of the

anterior segment with optical coherence tomography.

Investigative Ophthalmology & Visual Sciences 1993;

34:942.

[5] Michelson G, Schmauss B. Two-dimensional mapping

of the perfusion of the retina and optic nerve head. Br J

Ophthalmol 1995;79:1126–32.

[6] Hee M, Izatt J, Swanson E, et al. Optical coherence

tomography of the human retina. Arch Ophthalmol

1995;113:325–32.

[7] Michelson G, Schmauss B, Langhans M, et al. Princi-

ple, validity, and reliability of scanning laser doppler

flowmetry. J Glaucoma 1996;5:99–105.

[8] Hee M, Puliafito C, Duker J, et al. Topography of

diabetic macular edema with optical coherence tomog-

raphy. Ophthalmology 1998;105:360–70.

[9] Lobo C, Bernardes R, Santos F, Cunha-Vaz J. Mapping

retinal fluorescein leakage with confocal scanning laser

fluorometry of the human vitreous. Arch Ophthalmol

1999;117:631–7.

[10] Denninghoff KR, Smith MH, Hillman L. Retinal im-

aging techniques in diabetes. Diabetes Technol Ther

2000;2:111–3.

[11] Hood D. Assessing retinal function with the multifocal

technique. Prog Retin Eye Res 2000;19:607–46.

[12] Lobo C, Bernardes R, Cunha-Vaz J. Alterations of the

blood-retinal barrier and retinal thickness in preclinical

retinopathy in subjects with type 2 diabetes. Arch Oph-

thalmol 2000;118:1364–9.

[13] Konno S, Akiba J, Yoshida A. Retinal thickness mea-

surements with optical coherence tomography and the

scanning retinal thickness analyzer. Retina 2001;21:

57–61.

[14] Lobo C, Bernardes R, Faria de Abreu J, Cunha-Vaz J.

One-year follow-up of blood-retinal barrier and retinal

thickness alterations in patients with type 2 diabetes

mellitus and mild nonproliferative retinopathy. Arch

Ophthalmol 2001;119:1469–74.

[15] Bagga H, Greenfield DS, Feuer W, Knighton RW.

Scanning laser polarimetry with variable compensation

and optical coherence tomography in normal and glau-

comatous eyes. Am J Ophthalmol 2003;135(4):521–9.

[16] Miglior S, Guareschi M, Albe E, Gomarasca S, Vavas-

sori M, Orzalesi N. Detection of glaucomatous visual

field changes using the Moorfields regression analysis

of the Heidelberg retina tomograph. Am J Ophthalmol

2003;136(1):26–33.

[17] Lan YW, Henson DB, Kwartz AJ. The correlation be-

tween optic nerve head topographic measurements,

peripapillary nerve fiber layer thickness and visual field

indices in glaucoma. Br J Ophthalmol 2003;87(9):

1135–41.


[18] Schuman JS, Wollstein G, Farra T, Hertzmark E, Aydin

A, Fujimoto JG, et al. Comparison of optic nerve head

measurements obtained by optical coherence and con-

focal scanning laser ophthalmoscopy. Am J Ophthal-

mol 2003;136(2):404.

[19] Harris A, Dinn RB, Kagemann L, Rechtman E. A

review of methods for human retinal oximetry. Oph-

thalmic Surg Lasers Imag 2003;34(2):152–64.

[20] Pollack A, Aloni E, Nelson D, Grinvald A. Noninva-

sive imaging of blood flow and oximetry in diabetic

retinopathy patients, using a new retinal functional

imager [abstract]. The Israeli Society of Ophthalmol-

ogy 2001.

[21] Ofri R, Nelson D, Bonhoeffer T, Grinvald A. Nonin-

vasive imaging of blood flow and functional activation

of the cat retina using a new retinal function imager

[abstract]. The Israeli Society of Ophthalmology 2001.

[22] Nelson DA, Ofri R, Vanzetta I, Grinvald A. Dynamics

and regulation of blood flow rate in cat and human

retina [abstract]. Invest Ophthalmol Vis Sci 2002;43:

E-Abstract 843.

[23] Aloni EH, Pollack A, Grinvald A, Vanzetta I, Nelson

DA. Noninvasive imaging of retinal blood flow and

oximetry by a new retinal function imager [abstract].

Invest Ophthalmol Vis Sci 2002;43:E-Abstract 2552.

[24] Vanzetta I, Nelson DA, Bonhoeffer T, Ofri R, Grinvald

A. Novel intrinsic optical signals in feline and hu-

man retina evoked by photic stimulation [abstract].

Invest Ophthalmol Vis Sci 2002;43:E-Abstract 4363.

[25] Grinvald A, Lieke E, Frostig RD, Gilbert CD, Wiesel

TN. Functional architecture of cortex revealed by op-

tical imaging of intrinsic signals. Nature 1986;324:

361–4.

[26] Ts’o DY, Frostig RD, Lieke E, Grinvald A. Func-

tional organization of primate visual cortex revealed

by high resolution optical imaging. Science 1990;249:

417–20.

[27] Frostig RD, Lieke EE, Ts’o DY, Grinvald A. Cortical

functional architecture and local coupling between

neuronal activity and the microcirculation revealed

by in vivo high-resolution optical imaging of intrinsic

signals. Proc Natl Acad Sci USA 1990;87:6082–6.

[28] Grinvald A, ShohamD, Shmuel A, Glaser ED, Vanzetta

I, Shtoyerman E, et al. In vivo optical imaging of corti-

cal architecture and dynamics. In: Windhorst U, Johans-

son H, editors. Modern techniques in neuroscience

research. Umea, Sweden: Springer; 1999. p. 893–969.

[29] Mountcastle VB. Modality and topographic properties

of single neurons of cat’s somatic sensory cortex.

J Neurophysiol 1957;20:408–34.

[30] Hubel DH, Wiesel TN. Receptive fields, binocular

interactions and functional architecture in the cat’s vi-

sual cortex. J Physiol 1962;160:106–54.

[31] Hubel DH, Wiesel TN. Ferrier lecture. Functional ar-

chitecture of macaque monkey visual cortex. Proc R

Soc Lond 1977;198:1–59.

[32] Tasaki I, Watanabe A, Sandlin R, Carnay L. Changes

in fluorescence, turbidity and birefringence associated

with nerve excitation. Proc Natl Acad Sci USA 1968;

61:883–8.

[33] Salzberg BM, Davila HV, Cohen LB. Optical recording

of impulses in individual neurons of an invertebrate

central nervous system. Nature 1973;246:508–9.

[34] Grinvald A, Salzberg BM, Cohen LB. Simultaneous

recording from several neurons in an invertebrate cen-

tral nervous system. Nature 1977;268:140–2.

[35] Grinvald A, Anglister L, Freeman JA, Hildesheim R,

Manker A. Real time optical imaging of naturally

evoked electrical activity in the intact frog brain. Na-

ture 1984;308:848–50.

[36] Kelin D. On cytochrome, a respiratory pigment, com-

mon to animals, yeast, and higher plants. Proc R Soc

Lond B Biol Sci 1925;98:312–39.

[37] Millikan GA. Experiments on muscle hemoglobin in

vivo; the instantaneous measurement of muscle metabo-

lism. Proc R Soc Lond B Biol Sci 1937;123:218–41.

[38] Hill DK, Keynes RD. Opacity changes in stimulated

nerve. J Physiol 1949;108:278–81.

[39] Chance B, Cohen P, Jobsis F, Schoener B. Intracellular

oxidation-reduction states in vivo. Science 1962;137:

499–508.

[40] Jobsis FF. Noninvasive, infrared monitoring of cerebral

and myocardial oxygen sufficiency and circulatory

parameters. Science 1977;198:1264–6.

[41] Cohen LB. Changes in neuron structure during action

potential propagation and synaptic transmission. Phys-

iol Rev 1973;53:373–418.

[42] Roy C, Sherrington C. On the regulation of the blood

supply of the brain. J Physiol 1890;11:85–108.

[43] Raichle ME, Martin WRW, Herscovitz P, Minton MA,

Markham JJ. Brain blood flow measured with intrave-

nous H2(15)0. II. Implementation and validation.

J Nucl Med 1983;24(9):790–8.

[44] Fox PT, Mintun MA, Raichle ME, Miezin FM, Allman

JM, van Essen DC. Mapping human visual cortex

with positron emission tomography. Nature 1986;323:

806–9.

[45] Ogawa S, Tank DW, Menon R, Ellermann JM, Kim

SG, Merkle H, et al. Intrinsic signal changes accompa-

nying sensory stimulation: functional brain mapping

with magnetic resonance imaging. Proc Natl Acad

Sci USA 1992;89:5951–5.

[46] Malonek D, Grinvald A. The imaging spectroscope

reveals the interaction between electrical activity and

cortical microcirculation: implication for functional

brain imaging. Science 1996;272:551–4.

[47] Vanzetta I, Grinvald A. Cortical activity-dependent oxi-

dative metabolism revealed by direct oxygen tension

measurements: implications for functional brain imag-

ing. Science 1999;286:1555–8.

[48] Vanzetta I, Grinvald A. Evidence and lack of evidence

for the initial dip in the anesthetized rat: implications

for human functional brain imaging. Neuroimage

2001;13:959–67.

[49] Shtoyerman E, Arieli A, Slovin H, Vanzetta I, Grinvald

A. Long-term optical imaging and spectroscopy reveal


mechanisms underlying the intrinsic signal and stabil-

ity of cortical maps in V1 OF behaving monkeys.

J Neurosci 2000;20:8111–21.

[50] Cohen LB, Keynes RD, Hille B. Light scattering and

birefringence changes during nerve activity. Nature

1968;218:438–41.

[51] Grinvald A, Manker A, Segal M. Visualization of the

spread of electrical activity in rat hippocampal slices

by voltage-sensitive optical probes. J Physiol 1982;

333:269–91.

[52] Bonhoeffer T, Grinvald A. Iso-orientation domains in

cat visual cortex are arranged in pinwheel-like patterns.

Nature 1991;353:429–31.

[53] Malonek D, Tootell RBH, Grinvald A. Optical imaging

reveals the functional architecture of neurons process-

ing shape and motion in owl monkey area MT. Proc R

Soc Lond 1994;258:109–19.

[54] Shmuel A, Grinvald A. Functional organization for

direction of motion and its relation to orientation se-

lectivity in cat area 18. J Neurosci 1996;16:6945–64.

[55] Weliky M, Bosking WH, Fitzpatrick D. A systematic

map of direction preference in primary visual cortex.

Nature 1996;379:725–8.

[56] Shoham D, Grinvald A. Visualization of the hand rep-

resentation in macaque and human area S-I using

intrinsic signal optical imaging. J Neurosci 2001;21:

6820–35.

[57] Ghose GM, Tso DY. Form processing modules in pri-

mate area V4. J Neurosci 1997;77(4):2191–6.

[58] Wang G, Tanaka K, Tanifuji M. Optical imaging of

functional organization in the monkey inferotemporal

cortex. Science 1996;272:1665–8.

[59] Bosking WH, Zhang Y, Schofield B, Fitzpatrick D.

Orientation selectivity and the arrangement of horizon-

tal connections in tree shrew striate cortex. J Neurosci

1997;17:2112–27.

[60] Basole A, White LE, Fitzpatrick D. Mapping multiple

features in the population response of visual cortex.

Nature 2003;424:986–90.

[61] Masino SA, Frostig RD. Quantitative long-term imag-

ing of the functional representation of a whisker in rat

barrel cortex. Proc Natl Acad Sci USA 1996;93(10):

4942–7.

[62] Bakin JS, Kwon MC, Masino SA, Weinberger NM,

Frostig RD. Suprathreshold auditory cortex activation

visualized by intrinsic signal optical imaging. Cereb

Cortex 1996;6:120–30.

[63] Hess A, Scheich H. Optical and FDG mapping of fre-

quency-specific activity in auditory cortex. Neurore-

port 1996;7:2643–7.

[64] Harrison RV, Harel N, Panesar J, Mount RJ. Blood

capillary distribution correlates with hemodynamic-

based functional imaging in cerebral cortex. Cereb

Cortex 2002;12:225–33.

[65] Rubin BD, Katz LC. Optical imaging of odorant rep-

resentations in the mammalian olfactory bulb. Neuron

1999;23:499–511.

[66] Gratton G, Corballis PM, Cho E, Fabiani M, Hood DC.

Shades of gray matter: noninvasive optical images of

human brain responses during visual stimulation. Psy-

chophysiology 1995;32(5):505–9.

[67] Hoshi Y, Tamura M. Dynamic multichannel near-infra-

red optical imaging of human brain activity. J Appl

Physiol 1993;75:1842–6.

[68] Kato T, Kamei A, Takashima S, Ozaki T. Human vi-

sual cortical function during photic stimulation moni-

toring by means of near-infrared spectroscopy. J Cereb

Blood Flow Metab 1993;13:516–20.

[69] Livingstone MS, Hubel DH. Anatomy and physiology

of a color system in the primate visual-cortex. J Neuro-

sci 1984;4(1):309–56.

[70] Bartfeld E, Grinvald A. Relationships between orien-

tation preference pinwheels, cytochrome oxidase blobs

and ocular dominance columns in primate striate cor-

tex. Proc Natl Acad Sci USA 1992;89:11905–9.

[71] Vanzetta I, Slovin H, Omer DB, Grinvald A. Columnar

resolution of blood volume and oximetry functional

maps in the behaving monkey; implications for fMRI.

Neuron 2004, in press.


Electrophysiologic imaging of retinal and optic nerve

damage: the multifocal technique

Donald C. Hood, PhD

Department of Psychology, Columbia University, Schermerhorn Hall, Room 416, MC 5501, 1190 Amsterdam Avenue,

New York, NY 10027, USA

Traditional electrophysiologic tests of vision in- multifocal equipment. Before summarizing how

volve stimulation of relatively large areas of the ret-

ina [1]. For the standard electroretinogram (ERG)

and flash visual evoked potential (VEP) tests, the

entire retina is illuminated, whereas for the pattern

ERG and VEP tests, the stimulus typically exceeds

15 degrees in diameter. The size of the stimuli used for

these tests presents a problem if the clinician is

interested in the local topography of the damage to

the retina or optic nerve. Although ERG and VEP

responses can be elicited with relatively small stimuli,

each retinal area must be tested separately. The time

required to obtain multiple responses to construct a

topographical map would be prohibitive. The multi-

focal technique was devised by Sutter to solve this

problem [2]. Using the multifocal ERG (mfERG) and

VEP (mfVEP) techniques, local responses are

recorded simultaneously from many regions of the

visual field. This article provides an introduction to

these techniques.

The multifocal electroretinogram

Although the mfERG is a relatively new tech-

nique, it is used widely to diagnose and study ret-

inal diseases [3]. More than 200 articles, most of

them clinical in nature, have been published in the

last 5 years, and hundreds of centers worldwide have


doi:10.1016/S0896-1549(03)00101-9

This article was supported by grants EY02115

and EY09706 from the National Eye Institute, Bethesda,

Maryland.


the mfERG is used in the clinic, the basics of the

technique should be reviewed. The following sec-

tion provides an overview of how the mfERG is re-

corded. For more details, the reader is referred to the

article by Sutter and Tran [2], the International Society

for Clinical Electrophysiology of Vision (ISCEV)

guidelines [4], the special issue of Documenta Oph-

thalomologica published in 2000 (volume 100) on the

multifocal technique, and the review by Hood [3].

Method for recording the multifocal

electroretinogram

The display

Typically, the multifocal stimulus is displayed

on a computer monitor, although LED displays have

also been employed. The display contains an array of

hexagons; the most commonly used displays contain

either 61 or 103 hexagons (Fig. 1A). The hexagons

are usually scaled to produce local responses of

approximately equal amplitude in control subjects

[2]. The retinal size of the display varies across

laboratories and clinics but is generally 40 to

50 degrees in diameter. The ISCEV guidelines [4]

recommend that the luminance of the white hexa-

gons be set to 100 to 200 cd/m2 and the luminance of

the black hexagons to the lowest value available.

When the recording starts, the display appears

to flicker owing to the fact that each hexagon goes

through a pseudorandom sequence of black and white

presentations (Fig. 1B,C). In the most commonly used

software (VERIS; EDI, San Mateo, California), this

sequence is called an m-sequence [2].

s reserved.

Fig. 1. (A) The mfERG display. (B) The mfERG display as it might appear at any moment in time. (C) Examples of the

sequence of events at two locations. (D) A schematic of the continuous ERG signal. (E) mfERG responses, which are the

first-order correlations between the ERG signal (D) and the light sequence (C) of each hexagon. (Modified from Hood DC.

Assessing retinal function with the multifocal technique. Prog Retin Eye Res 2000;19:607–46; with permission).

D.C. Hood / Ophthalmol Clin N Am 17 (2004) 69–8870

The recording

Using the same electrodes and amplifiers em-

ployed for standard full-field ERG recording, a single

continuous ERG record is obtained (Fig. 1D), usu-

ally with the pupil dilated. The continuous record

is typically 4 to 8 minutes in length and is obtained

in 15- to 30-second segments for the subject’s

comfort. The ISCEV guidelines [4] recommend a

low-frequency amplifier cutoff of 300 Hz and a

high-frequency cutoff of 3 or 10 Hz. The 10-Hz set-

ting will produce changes in the waveforms but is

used more typically than the 3-Hz setting, because it

produces more stable recordings.

Deriving and presenting the multifocal

electroretinogram responses

The mfERG responses for stimulation of a control

subject’s right eye are shown in Fig. 1E. These

responses are derived from the single continuous

ERG record (Fig. 1D). Technically, the mfERG re-

sponses are derived as the first-order kernels of the

cross correlation between the stimulation sequence

and the continuously recorded ERG [2,3,5–7].

The responses in Fig. 1E are positioned so that

they do not overlap; therefore, the scaling is not linear

as the isodegree circles in Figs. 1A and 1E indicate.

Frequently, it is useful to sum or average the responses

within various regions of the display. Fig. 2B shows

the responses from Fig. 1E summed within rings

around fixation (Fig. 2A). The summed responses

become larger with eccentricity, because progres-

sively larger areas of the retina are stimulated. To

take the area of stimulation into consideration, for

each ring, the summed response (Fig. 2B) is scaled by

the total area of the hexagons in that ring. The

resulting response (Fig. 2C) is expressed as the

response amplitude per unit area (nV/deg2) or re-

sponse density. Response density is largest in the

fovea [2]. The software available to analyze the

mfERG allows the combination of responses from

any arbitrary grouping of hexagons.

Comparison of the multifocal electroretinogram with

the full-field test

As typically recorded, the mfERG is a cone-driven

response from the central 20 to 30 degrees (radius) of

the retina (see Fig. 1A). The temporal rate of stimu-

lation combined with the light levels employed

D.C. Hood / Ophthalmol Clin N Am 17 (2004) 69–88 71

ensures that the rods do not contribute to the response

except under unusual circumstances [8,9].

The mfERG shows an initial negative component

(N1) followed by a positive component (P1) and

then a second negative component (N2) (Fig. 2C).

Although the N1 and P1 components of the mfERG

resemble the a- and b-waves of the full-field flash

ERG, the waveforms differ [3,10]. The differences

between the waveforms of the mfERG and the full-

Fig. 2. (A) The rings used for the analysis in panels B and C. (B) T

(C) The mfERG responses in panel B divided by the area within e

Odel JG, Chen CS, Winn BJ. The multifocal electroretinogram (E

J Neuroophthalmol 2003;23:225–35.).

field ERG have been attributed to differences in the

methods of light stimulation and the methods of

deriving the responses [10]. In fact, unlike the full-

field ERG, the mfERG is not a response at all but a

mathematical extraction; therefore, the components

of the mfERG should never be referred to as a- and

b-waves. Despite these differences, the N1 response

of the human mfERG (Fig. 2C) is composed of

the same components as the a-wave of the full-field

he mfERG responses summed by rings and expressed in mV.ach ring and expressed in nV/deg2. (Modified from Hood D,

RG): applications and limitations in neuro-ophthalmology.


ERG, and the P1 response is composed of the

same components as the positive waves (b-wave and

oscillatory potentials [OPs]) [10]. If the rate of stimu-

lation of the multifocal sequence is slowed by inter-

leaving seven or more blank frames (93 ms or more),

the waveform of the human mfERG closely resembles

the waveform of the full-field ERG [10].

Topographical measure of diseases of the outer retina

The mfERG provides a topographical indicator

of the health of the outer retina. The standard hu-

man mfERG is largely shaped by bipolar cell activ-

ity, with smaller contributions from the photoreceptor

and inner retinal (eg, amacrine and ganglion) cells

[11,12]. Fig. 3 illustrates how the cells of the retina

contribute to determine the shape of the human

mfERG [11]. A similar analysis of the mfERG can

be elicited by a slower stimulation rate [13]. When the

standard mfERG paradigm is used, the inner retina

makes a relatively small contribution to the wave-

forms. In fact, removing the activity of the ganglion

cells has a relatively minor effect on the standard

human mfERG [14]. In the extreme, subtle changes

Fig. 3. A model illustrating contributions of the different retinal ce

Frishman LJ, Saszik S, Viswanathan S. Retinal origins of the prim

Invest Ophthalmol Vis Sci 2002;43:1673–85; with permission.)

in the waveform can be observed. For a disease to

decrease the amplitude of the mfERG, the cone

photoreceptors or cone-driven bipolar cells must be

abnormal. The mfERG, like the standard full-field

ERG, provides a measure of the health of the outer

retina (ie, cone photoreceptors and bipolar cells).

For diseases of the outer retina, the mfERG pro-

vides a topographical picture of retinal damage. Fig. 4

compares the recordings obtained with the standard

(ISCEV), full-field, clinical ERG protocol with the

mfERG for a patient with retinitis pigmentosa. For the

full-field ERG (left column), the 30-Hz flicker stimu-

lus produces a response of less than 300 nV (bottom of

left column), whereas the responses to the other

stimuli are not detectable. The ERG indicates that

the retina is still producing a small cone-driven signal.

The mfERG (Fig. 4, right column) shows that this

signal is coming almost entirely from the central

5 degrees. For this patient, the full-field ERG appears

to be composed of local responses of similar wave-

forms. In other patients, the waveform of the local

responses can differ markedly, which may result in

a full-field ERG with a waveform that is not repre-

sentative of any of the local responses [3,15].

ll types to the mfERG waveform. (Modified from Hood DC,

ate multifocal ERG: implications for the human response.

Fig. 4. The full-field ERG records from a standard clinical protocol (left column) and the mfERG records (right column) are

shown for a patient with retinitis pigmentosa. (From Hood DC. Assessing retinal function with the multifocal technique. Prog

Retin Eye Res 2000;19:607–46; with permission).


The second-order kernel

Technically, the responses discussed thus far are

first-order kernels. When commercially available soft-

ware is used, it is possible to derive higher order

kernels. In general, the only higher order kernel

typically reported is the first slice of the second-order

kernel. Fig. 5 provides a simple way to understand the

meaning of the second-order kernel. The schematic

shows that, for any given hexagon, the presentation

of a flash of light can be preceded in the previous

frame by either a flash or darkness. If the response

Fig. 5. A schematic illustrating how the second-order kernel is derived. See text for details. (Modified from Hood D, Odel JG,

Chen CS, Winn BJ. The multifocal electroretinogram (ERG): applications and limitations in neuro-ophthalmology. J Neu-

roophthalmol 2003;23:225–35.)


after the flash differs from the response after the blank

frame, there is a second-order kernel and this kernel is

equal to the difference between the responses [3,5].

The second-order kernel is a measure of how the

mfERG response is affected by a preceding flash. It

is not a response of a particular subset of retinal cells.

The integrity of the outer and inner retina is im-

portant for a normal second-order kernel. In mon-

keys, blocking action potentials generated by the

ganglion cells and amacrine cells markedly reduce

but do not eliminate the second-order kernels [11,12].

In humans, damage to the ganglion cells does not

necessarily change the second-order kernels [14];

however, outer plexiform damage can eliminate the

second-order kernels in patients with degenerative

diseases of the receptors [3,16]. Consequently, a

diminished second-order kernel does not imply dam-

age to a particular set of cells.

Clinical uses of the multifocal electroretinogram

Although the mfERG is used in clinics world-

wide, many patients who could benefit from mfERG

are not tested. Other patients are tested using the

mfERG when full-field ERG tests would be as, if

not more, informative. The mfERG test is best re-

served for situations in which the full-field ERG is

normal or is likely to be normal or uninformative. The

examples described in this section and elsewhere [3,7]

illustrate such cases.

Fig. 6. (A) The OS (left panel) and OD (right panel) 24-2 HVFs of

and OD (right panel) mfERG responses for this patient. The vertic

respectively. Bold dark gray, thin black, and dashed light gray circle

OS (left panel) and OD (right panel) mfERG responses, expressed

between 5 and 15 degrees (black), and between 15 and 25 degrees (l

BJ. The multifocal electroretinogram (ERG): applications and lim

23:225–35.)

Establishing that a disease has an outer retinal origin

Frequently, the clinician must determine whether

a visual field defect is caused by damage to the outer

retina (ie, receptors and bipolar cells) or to the gan-

glion cells/optic nerve. In most cases, when the site of

damage is the outer retina, the fundus examination,

angiogram, or full-field ERG will be abnormal.

Nevertheless, in some circumstances, damage to the

outer retina can be missed by these tests. Fig. 6B

shows the mfERG responses from a patient with a

defect that would be hard to detect using the tradi-

tional full-field ERG or focal ERG. The fundus in this

patient appeared normal, and his visual acuity was

20/20. 24-2 Humphrey visual fields (HVF) (Fig. 6A)

showed paracentral ring scotomas in both eyes. The

records at the bottom of Fig. 6A show the average

response density within the three areas marked. The

amplitudes of the responses are reduced markedly in

the ring with the most severe field damage. The ques-

tion for the ophthalmologist was whether the field

defect was caused by glaucoma or a retinopathy. The

mfERG confirmed that the problem was in the outer

retina. It would be difficult to detect this defect using

the traditional full-field ERG or focal ERG [1,17].

Comparing the multifocal electroretinogram with

visual field topographies

The clinical utility of the mfERG can be enhanced

greatly by comparing the findings with the visual

fields from static automated perimetry. As is true for

a patient with paracentral scotomas. (B) The OS (left panel)

al and horizontal calibration bars indicate 100 nV and 60 ms,

s indicate radii of 5, 15, and 25 degrees, respectively. (C) The

as response density, for the area within 5 degrees (dark gray),

ight gray). (Modified from Hood D, Odel JG, Chen CS, Winn

itations in neuro-ophthalmology. J Neuroophthalmol 2003;


all clinical measures, variability can lead to a degree

of uncertainty in the interpretation of mfERG results;

however, much of this uncertainty disappears if the

mfERG and visual field topographies agree. To com-

pare the HVF with the mfERG topographies, iso-

degree contours can be added as shown by the circles

in Fig. 6.Althoughmore sophisticated procedures exist

for comparing the HVF with the mfERG [18], these

contours are sufficient for most clinical purposes. The

agreement between the mfERG and HVF topogra-

Fig. 7. (A) The 30-2 HVF of a patient with retinitis pigmentosa.

responses, expressed as response density (C) and as summed respo

and 15 degrees (black), and between 15 and 25 degrees (light gra

phies in Fig. 6 establishes the retinal origin of the

defect with a degree of certainty not possible with

tests that do not provide topographical information.

Differentiating among different retinal diseases

The specific pattern of amplitude and implicit

time abnormalities can help to distinguish different

retinal diseases. A large delay in the timing of the

mfERG is associated with damage to the receptors/

outer plexiform layer [3,15,16,19,20]. In diseases

(B) The mfERG responses for this patient. (C) The mfERG

nses (D) for the area within 5 degrees (dark gray), between 5

y).


such as retinitis pigmentosa and cone dystrophies,

the responses in regions of depressed sensitivity

sometimes show marked delays. Fig. 7 shows the

HVF field (panel A) and mfERG responses (panel B)

for a patient with retinitis pigmentosa. The responses

from the central regions have normal timing, whereas

the responses from the peripheral regions, where the

fields are depressed, show large delays. This differ-

ence can be seen more easily in panel C, where the

response densities (see Fig. 2) are shown for the three

regions indicated in panels A and B. Because the

central region represents a small proportion of the

total field, the full-field ERG will be dominated in

Fig. 8. (A) The 10-2 HVF of a patient after surgery to repair a ma

three-dimensional representation of the mfERG results. (D) The res

were averaged for regions within and outside of the defect. The d

this patient by the peripheral responses (summed

responses in Fig. 7D). In fact, this patient has a large

but delayed 30-Hz full-field flicker response. Inter-

estingly, although retinitis pigmentosa is a disease of

the receptors, the retina can still produce large but

delayed responses in some patients.

Although large delays in timing are a sign of

degenerative photoreceptor disease, local lesions to

all layers and damage to the inner nuclear layer

(eg, bipolars) are usually associated with smaller

responses that have minimal delays in timing. The

patient whose mfERG records appear in Fig. 8B

complained of a ‘‘C-shaped’’ field problem following

cular hole. (B) The mfERG responses for this patient. (C) A

ponses from the affected eye (black) and the other eye (gray)

ashed curve is the black curve scaled in amplitude.


surgery to repair a macular hole. Her 10-2 HVF

(Fig. 8A) matched the subjective complaints. The

depressed amplitudes in the mfERG responses seen

in Fig. 8B and the three-dimensional representation in

Fig. 8C closely correspond to the field. This finding

suggests that the damage is at or before the bipolar

cells. Damage to the nerve fiber layer would not

produce these local losses in amplitude. Fig. 8D shows

the average responses from the good eye (gray) and the

affected eye (black) for a region within the defect

(upper traces), as well as for a region outside the defect

(lower traces). The amplitude is reduced with little

change in implicit time. The dashed curve represents

the black record scaled in amplitude. In Fig. 8D,

in addition to a reduction in amplitude, there is a subtle

change in waveform, with N2 relatively reduced in

amplitude. The second-order kernel (not shown) is

missing in the affected region, accounting in part for

this change in waveform [3,5]. Local lesions are often

easier to see in the second-order kernel [21,22].

Diabetic retinopathy can affect the inner and outer

retinal layers. Consequently, the mfERG from pa-

tients with diabetic retinopathy can show a range of

changes [23,24]. Some patients have normal mfERG

responses in regions of poor sensitivity, consistent

with inner retinal damage. Other patients show large

but delayed responses in such regions, consistent with

damage in the outer retina, probably the outer plexi-

form layer [3].

Other clinical uses

There are other less common uses of the mfERG

in addition to establishing the retinal locus of a

disease process and differentiating among diseases.

In a review of cases seen by neuro-ophthalmologists,

the author and his colleagues [7] found that the

mfERG could be used to rule out functional (non-

organic) causes, to follow the progression of a disease,

and to distinguish between two pre-existing condi-

tions that could have caused a visual field defect. In

addition, mfERG has been used to examine patients

before and after treatment for retinal detachment [25],

laser treatment for macular edema secondary to dia-

betic retinopathy [23], and surgery for macular holes

[26]. Other uses of the mfERG technique will most

likely be found as new interventions are developed.

The mfERG should be particularly useful in situations

in which localized changes are expected, such as

experimental retinal transplant procedures and local

drug injections.

The need for normative values

In all types of electrophysiologic testing, each

clinic should establish its own normative values.

For reference purposes, normative values have been

published [27–31].

Detecting ganglion cell damage with the multifocal

electroretinogram

The effectiveness of the mfERG for detecting local

ganglion cell damage in humans is currently under

debate. Although the literature contains various con-

tradictory findings, the evidence is relatively clear on

some points. First, a component generated at the optic

nerve head seems to reflect local ganglion cell activity.

Sutter and Bearse [32] first identified this component

in the human mfERG, calling it the optic nerve head

component (ONHC). Second, a component similar to

the ONHC has been identified in the monkey mfERG

[33]. When action potentials are blocked in the

monkey’s eye, this ONHC disappears, as does a

high-frequency component that resembles oscillatory

potentials [33–35]. Because only the ganglion cells

and a few classes of amacrine cells fire action po-

tentials, the ONHC undoubtedly depends on action

potentials produced by the ganglion cells. On the

other hand, it is not clear whether the high-frequency

component depends on ganglion cell or amacrine cell

activity. The ONHC and the high-frequency compo-

nent are larger in the monkey mfERG than in the

human mfERG [33]. Because these inner retinal

contributions are large in the monkey, it is relatively

easy to see the effects of glaucomatous damage in

monkey models [36,37]. Some patients with glau-

coma show changes in the standard mfERG, and these

changes can often be related to measures of the OHNC

or high-frequency component [14,38–42]. Neverthe-

less, attempts to detect glaucomatous damage with the

standard mfERG recordings show relatively poor

sensitivity and specificity [14,40,41,43]. The rela-

tively small ONHC in humans can be enhanced with

specialized paradigms of mfERG stimulation (eg,

the global flash paradigm [44–46]) or methods of

analysis [32]. Even with these methods for enhancing

the contributions of the inner retina, the results remain

controversial. Although, in a few patients, clear evi-

dence of local damage has been reported [42,45,47],

in general, the results published to date have been

disappointing [47,48]. It remains unclear whether the

mfERG, combined with a global flash paradigm, can

detect early damage. If the results of future studies are

more encouraging, the mfERG technique should be

compared with other objective tests of ganglion cell

function, such as the pattern ERG, the photopic

negative response, and the multifocal VEP [49]. For

now, the mfERG cannot be considered a useful clini-

cal tool for studying ganglion cell damage.

l Clin N Am 17 (2004) 69–88 79

Common mistakes

Two common mistakes are worth mentioning.

First, three-dimensional representations, such as that

shown in Fig. 8C, should never be presented without

the accompanying response array (Fig. 8B). Records

without any signal can produce three-dimensional

plots that have a peak in the center, that is, noise

by itself can appear as if there were a foveal response

present [4,7]. Second, eccentric fixation can mimic a

macular problem [7]. This effect is particularly trou-

blesome in the patient who has a small central loss of

vision that could be caused by optic atrophy or outer

retinal disease. Optic atrophy, coupled with eccentric

fixation, can mimic the mfERG recorded when an

outer retinal defect is present. In cases of central loss,

it is important to have a way of determining where

the patient is looking [7].

Summary

The mfERG is a useful clinical tool [3]. It is par-

ticularly valuable for distinguishing damage to the

outer retina (choroidal, receptor, and bipolar damage)

from damage to the ganglion cells and optic nerve

when the full-field ERG is normal. In many cases, the

changes in amplitude and waveform can help identify

the particular level at which the retina is affected.

Technical problems must be overcome to obtain use-

ful mfERG recordings. If the clinician is already

recording high-quality full-field ERGs, recording

high-quality mfERGs should not be difficult. Never-

theless, the difficulties involved in full-field ERG

testing are also encountered in the recording and the

analysis of mfERG, and other difficulties unique to

the multifocal technique arise. In general, mfERG

testing is best done at centers with an electrophysiolo-

gist trained in ERG and multifocal technology.

D.C. Hood / Ophthalmo

The multifocal visual evoked potential

The VEP is a gross electrical potential generated

by the cells in the occipital cortex. VEP responses can

be recorded simultaneously from many regions of

the visual field using the mfVEP technique introduced

by Baseler and colleagues [50,51].

Method for recording the multifocal visual evoked

potential

The VEP signal is recorded using the same elec-

trodes and amplifiers employed for conventional VEP

recording. There are critical differences in the display,

the method of stimulation, and the analysis of the

raw records. This section provides basic information

on how the mfVEP is recorded. For more details, the

reader is referred to the review by Hood and Green-

stein [52].

The display

The mfVEP can be recorded with a wide range

of displays and paradigms. Most of the mfVEP work

to date has been done using pattern reversal stimula-

tion and the display in Fig. 9A, which was introduced

by Baseler and colleagues [50,51]. The dartboard dis-

play contains 60 sectors, each containing 16 checks,

8 black and 8 white. The sectors and the checks are

scaled based on cortical magnification. The retinal

size of the display varies across laboratories but is

generally between 30 and 60 degrees in diameter. For

the recordings presented herein, the diameter was

44.5 degrees. Although there are no ISCEV guidelines

for mfVEP recording, typically, the luminance of the

white hexagon is set between 50 and 200 cd/m2 and

the black hexagon to the lowest value available.

The recording

A single continuous VEP (EEG) record is ob-

tained with electrodes over the occipital region. In

general, the mfVEP is recorded with two midline

electrodes, which serve as the active and the reference

electrodes. A third electrode, the ground, is placed on

the forehead or the ear (the so-called ‘‘bipolar record-

ing’’). Klistorner and Graham [53] suggested record-

ing from four channels with two additional electrodes

placed lateral to the inion. Hood and colleagues

[52,54] use three ‘‘active’’ electrodes, one placed

4 cm above the inion and two placed 4 cm lateral to

the inion on each side. Each active electrode is

referenced to the inion, providing three channels of

recording. This technique effectively provides six

channels of records, because the recordings that

would result from three additional channels can be

derived with software (Fig. 9B). Fig. 9C shows the

mfVEP responses (midline channel) for a control

subject. The best responses, defined in terms of a

signal-to-noise ratio from all six channels, are shown

in Fig. 9D. For this individual, the additional channels

improved the recordings from the horizontal meridian

(black ellipses) and the fovea (gray ellipses), as

suggested by Klistorner and Graham [53].

Displaying the multifocal visual evoked potential

responses

The mfVEP responses are usually displayed as

an array shown in Figs. 9C and D. The spatial scale

Fig. 9. (A) The display employed for the mfVEP recordings. (B) The electrode locations employed for the mfVEP recordings.

(C) Array of mfVEP responses obtained from the midline electrode placements (A–D in panel B). (D) Array of mfVEP

responses obtained from the recording channels that produce the best responses, as defined by the highest signal-to-noise ratio.


for this array is not linear as a comparison of the

isodegree circles in Figs. 9C and D indicates. The

responses in the array are positioned arbitrarily so that

they do not overlap. Within the central 2.6 degrees

(5.2 degrees in diameter), there are 12 responses,

whereas a region that is considerably larger than the

entire central 2.6 degrees produces each of the

responses in the outer ring.

As is true for the mfERG, each of the individ-

ual mfVEP responses in the array is not, technically,

a ‘‘response’’; rather, each response is derived via a

correlation between the stimulation sequence of a

particular sector and the overall single continuous

VEP recording.

Comparison of the multifocal visual evoked potential

responses from two eyes

Considerable intersubject variability occurs in the

amplitudes and waveforms of the mfVEP responses.

This variability occurs owing to individual differences

in the location and folding of the visual cortex [18,52].

Fig. 10 shows the mean responses obtained with

monocular stimulation of the right (black) and left

(grey) eyes of 30 normal control subjects. These

responses are nearly identical because they are gen-

erated in the same cortical regions. There are two

exceptions. First, there is a small amplitude asymme-

try along the horizontal meridian. Second, there is a

Fig. 10. The averaged mfVEP responses for 30 normal control subjects.

Fig. 11. A comparison of the conventional pattern VEP

(pVEP) and the mfVEP. Responses from the upper (dashed)

and lower (solid) hemifields are shown. (Modified from

Fortune B, Hood DC. Conventional pattern-reversal VEPs

are not equivalent to summed multifocal VEPs. Invest

Ophthalmol Vis Sci 2003;44:1364–75; with permission.)


small interocular latency difference of about 4 or 5 ms

across the midline. This small difference can be seen

in the insets in Fig. 10. The left eye leads in the left

visual field, and the right eye leads in the right visual

field. The article by Hood and Greenstein [52] pro-

vides a discussion of the reasons for these differences.

Comparison of the multifocal visual evoked potential

with the conventional test

The major components of the mfVEP resemble

those of the conventional VEP elicited by pattern

reversal stimuli. In Fig. 11, the conventional pattern

VEPs recorded with stimulation of the upper and

lower hemifields are compared with the summed

mfVEPs (from the midline channel) for the same

hemifields. In the mfVEP, there is an initial negative

component around 65 ms followed by a prominent

positive component (C2) around 95 ms, analogous

to the N75 and P100 of the conventional pattern VEP

[55]. Nevertheless, the mfVEP responses differ from

the pattern VEP. The pattern VEP from the upper

hemifield is often a complex waveform, whereas the

mfVEP responses above and below the midline are

reversed in polarity. As noted by Baseler et al [50],

the reversal in polarity of the mfVEP response sug-

gests that the primary generator of the mfVEP is in the


calcarine fissure. Fortune and Hood [55] speculated

about the causes of the differences between the

mfVEP and pattern VEP. They suggested that the

mfVEP is largely generated in V1 (striate cortex),

whereas the conventional VEP contains a relatively

larger extrastriate contribution. Recently, Zhang and

Hood [56] extracted a large principal component

of the mfVEP that appears to be generated in V1.

Topographical measure of diseases of the ganglion

cells and optic nerve

Local damage to the ganglion cells and optic

nerve can be detected with the mfVEP. To date, two

groups have developed quantitative techniques for

comparing mfVEP abnormalities with visual defects

measured with static automated perimetry [18,52–

54,57–62]. These techniques are more elaborate

Fig. 12. (A) The 24-2 HVF total deviation probability plot from th

the left (red) and right (blue) eyes. The red, blue, and green circles i

respectively. (C) The probability plots for the mfVEPs in panel B. S

The multifocal VEP and ganglion cell damage: applications and

2003;22:201–51; with permission.)

than those used in mfERG. To illustrate the problem,

consider the patient whose 24-2 HVF (probability

plot) is shown in Fig. 12A. This patient had unilat-

eral glaucomatous damage in the left eye; the HVF

for the right eye was normal. The defects in the left

eye are circled in gray and black. The mfVEP re-

sponses obtained from the patient’s left eye (red)

and right eye (blue) are shown in Fig. 12B. The iso-

degree contours are shown for the HVF and the

mfVEP responses.

To determine which of the responses from the

left eye (red records in Fig. 12B) were abnormal,

mfVEP probability plots analogous to the HVF prob-

ability plot in Fig. 12A were developed. Monocular

mfVEP probability plots (left two panels in Fig. 12C)

were obtained by comparing the patient’s mfVEPs

with the mfVEPs from the left and right eyes of a

group of control subjects [52]. Each square is the

e left eye of a patient with glaucoma. (B) The mfVEPs from

n panels A and B indicate radii of 2.6, 9.8, and 22.2 degrees,

ee text for details. (Modified from Hood DC, Greenstein VC.

limitations for the study of glaucoma. Prog Retin Eye Res


physical center of one of the sectors in the mfVEP

display (see Fig. 9A). For each sector, the amplitude

of the patient’s mfVEP was determined and compared

with values for a group of control subjects [52,54,

61,62]. A colored square indicates that the mfVEP

was statistically significant at the 5% (desaturated

color) or 1% (saturated color) level. The color indi-

cates whether the left (red) or right (blue) eye was

significantly smaller than normal.

The HVF (Fig. 12A) and mfVEP (Fig. 12C) pro-

bability plots are on the same linear scale. A direct

comparison can be made. To aid in this comparison,

the black and gray ellipses from Fig. 12A are overlaid

onto Fig. 12C. The mfVEP confirms the HVF defect

within the black ellipse but not the defect within the

gray ellipse.

In some patients, especially those with unilateral

damage, an interocular comparison of the mfVEPs

is a more sensitive indicator of damage [52,59] than

is monocular comparison with the control group. To

obtain the interocular mfVEP plot in Fig. 12C (right

hand panel), the ratio of the mfVEP amplitudes of

the two eyes was measured for each sector of the

display and compared with the ratios from a group

of controls [18,52,54,57,58]. The result was coded

as in the monocular fields. The defect within the

gray ellipse was still not apparent, but an arcuate

defect was detected in the lower field that was not

present in the HVF. Subsequent tests confirmed that

this defect was real. Hood and Greenstein [52] have

Fig. 13. (A) The 24-2 HVF total deviation probability plot from

mfVEPs from the left (gray) and right (black) eyes. (Modified from

potential (VEP): applications and limitations in neuro-ophthalmolo

provided a review of the derivation and use of

monocular and interocular probability plots.

Clinical uses of the multifocal visual evoked potential

Although the clinical role of mfVEP is still being

defined, five uses can be identified.

Diagnosing and following optic neuritis/multiple

sclerosis

Optic neuritis is a clinical syndrome characterized

by an acute unilateral loss of vision accompanied by

an afferent pupillary defect. It is often one of the first

signs of multiple sclerosis. Typically, the optic neu-

ritis partially or completely resolves within 3 months.

In fact, some patients have reasonably normal visual

fields after recovery. In some of these cases, the pa-

tients complain of ‘‘hazy’’ or ‘‘fuzzy’’ vision in parts

of the visual field, although the HVF can appear

normal in these regions. After recovery, the mfVEP

shows local delays, presumably in the regions cor-

responding to the portions of the optic nerve that

were demyelinated.

The mfVEP records in Fig. 13B show the range

of findings that can be observed in a patient sustaining

an attack of optic neuritis [63–65]. In this case, the

HVF probability plot (Fig. 13A) shows a paracentral

defect. The mfVEP (Fig. 13B) shows that, outside the

region of the defect on the HVF (black ellipse), there

are regions with delayed mfVEP responses (asterisks)

a patient with optic neuritis and multiple sclerosis. (B) The

Hood D, Odel JG, Winn BJ. The multifocal visual evoked

gy. J Neuroophthalmol 2003;23(3):225–35.)


and regions with reasonably normal mfVEP responses

(plus signs). In fact, regions with delays can border

regions that have responses with normal amplitude

and latency; therefore, the mfVEP can detect local

demyelinizaton [63].

The mfVEP is clearly superior to the HVF and

the conventional VEP for following patients with

optic neuritis and multiple sclerosis. Whether the

conventional pattern VEP is abnormal will depend

on the relative contributions of the normal and abnor-

mal regions. The conventional VEP is most likely to

miss problems in the upper field, which typically

contribute less to the overall VEP signal than does

the lower field [55]. Fig. 14 shows the visual field

probability plots (panel A) and mfVEP responses

(panel B) from a 45-year-old man who complained

of blurred vision in the upper field of the left eye. His

Fig. 14. (A) The 24-2 HVF total deviation probability plots from

mfVEPs from the left (gray) and right (black) eyes are shown.

conventional pattern VEP and HVFs (panel A) were

normal. The mfVEPs summed within each quadrant

are shown as the insets in panel B. The mfVEPs are

clearly delayed in the upper field. The diagnosis of

multiple sclerosis was confirmed on MRI studies

showing lesions in the left optic nerve.

Ruling out functional (nonorganic) causes

The conventional VEP has traditionally been used

to rule out functional or nonorganic causes of a visual

defect. Because multiple local responses are obtained,

the mfVEP is more effective than the conventional

VEP for this purpose. A local defect can be identified

on the mfVEP but missed totally on the conventional

VEP if the defect involves a small part of the total field

stimulated. In these cases, the diagnosis of a function-

al cause can be avoided. On the other hand, it is far

a patient with optic neuritis and multiple sclerosis. (B) The


easier to make the diagnosis of a functional cause

when large normal mfVEP responses are present in

regions of the field where the 24-2 HVF shows a pro-

found defect [66].

Identifying unreliable fields

The HVF test provides a measure of fixation errors

as well as false-positive and false-negative errors,

allowing the test to identify ‘‘unreliable fields.’’ Many

patients cannot produce reliable HVFs. For most of

these patients, the mfVEP provides an alternative.

Identifying questionable fields or fields that need

confirmation

In many cases, the ophthalmologist has insuffi-

cient or contradictory evidence, making it difficult to

diagnose the cause of a defect seen on the HVF. In

such cases, the mfVEP provides another topographi-

cal map that can help in the diagnosis [52].

Detecting glaucomatous damage

Ophthalmologists rely on visual fields from auto-

matic static perimetry such as the 24-2 HVF to detect

and to follow glaucomatous damage. Because glau-

coma can be controlled with medication and surgery,

early detection and therapeutic intervention are criti-

cal. Histologic studies on donor eyes have found that

HVF tests may not detect defects until substantial

ganglion cell damage has taken place [67–69]. This

finding has encouraged the search for new techniques

for identifying functional defects. The mfVEP has

been suggested as a test that may be superior to the

HVF [58].

Local ganglion cell damage secondary to glaucoma

can be detected with the mfVEP [18,52–54,57,58,

70–72]. In many cases, the mfVEP objective field loss

resembles the field loss measured with the subjective

HVF technique [52,58]. In other cases, the mfVEP

detects damage missed by the 24-2 HVF. The extent to

which the mfVEP will outperform standard white-on-

white perimetry remains an unanswered question [52].

Hood and Greenstein [52] have provided a theo-

retical framework for judging when the 24-2 HVF or

the mfVEP will be superior in detecting damage. They

start with the assumption that the mfVEP response is

composed of two components—signal and noise.

With this assumption, the evidence supports a simple

proportional relationship between the local loss of

ganglion cells and the amplitude of the signal in the

local mfVEP response [52,59], that is, losing one half

of the local sensitivity (�3 dB loss) results in a

decrease of the mfVEP signal by one half. Hood

et al [52,59] suggest that the decreases in local visual

field sensitivity and local mfVEP signal amplitude are

proportional to local ganglion cell loss.

Based on the Hood–Greenstein analysis and mea-

sures of the variability of the HVF and mfVEP, there

will be circumstances in which the mfVEP will detect

damage missed on the HVF, as well as circumstances

when the reverse will be true. The patient in Fig. 12

provides an example. The mfVEP detects a defect

missed by the 24-2 HVF (purple ellipse) in the lower

field, whereas the reverse is true for the defect in the

upper part of the 24-2 HVF (gray ellipse). The Hood–

Greenstein analysis suggests that the two tests will

often, but not always, agree. This theoretical analysis

also helps specify the conditions under which one

test will be better at detecting damage [52]. Recent

evidence supports this analysis [73].

Summary

Multifocal VEP technology is still in its infancy.

Although many aspects of the recording and analy-

sis will be improved with time, the mfVEP can be

used in the clinic today. It can help to identify

functional (nonorganic) problems, demyelinating dis-

eases, glaucomatous damage, and poor field takers.

Given the software and techniques needed to record

and interpret the results, mfVEP testing should be

performed in centers experienced with this technique.

References

[1] Holopigian K, Hood DC. Electrophysiology. Ophthal-

mol Clin N Am 2002;16:237–51.

[2] Sutter EE, Tran D. The field topography of ERG com-

ponents in man. I. The photopic luminance response.

Vision Res 1992;32:433–46.

[3] Hood DC. Assessing retinal function with the multi-

focal technique. Prog Retin Eye Res 2000;19:607–46.

[4] Marmor MF, Hood D, Keating D, Kondo M, Miyake

Y. Guidelines for basic multifocal electroretinography

(mfERG). Doc Ophthalmol 2003;106:105–15.

[5] Sutter E. The interpretation of multifocal binary ker-

nels. Doc Ophthalmol 2000;100:49–75.

[6] Keating D, Parks S, Evans A. Technical aspects of mul-

tifocal ERG recording. Doc Ophthalmol 2000;100:

77–98.

[7] Hood D, Odel JG, Chen CS, Winn BJ. The multifocal

electroretinogram (ERG): applications and limitations

in neuro-ophthalmology. J Neuroophthalmol 2003;23:

225–35.

[8] Wu S, Sutter EE. A topographic study of oscillatory

potentials in man. Vis Neurosci 1995;12:1013–25.

[9] Hood DC, Wladis EJ, Shady S, Holopigian K, Li J,


Seiple W. Multifocal rod electroretinograms. Invest

Ophthalmol Vis Sci 1998;39:1152–62.

[10] Hood DC, Seiple W, Holopigian K, Greenstein V. A

comparison of the components of the multifocal and

full-field ERGs. Vis Neurosci 1997;14:533–44.

[11] Hood DC, Frishman LJ, Saszik S, Viswanathan S. Reti-

nal origins of the primate multifocal ERG: implica-

tions for the human response. Invest Ophthalmol Vis

Sci 2002;43:1673–85.

[12] Hare WA, Ton H. Effects of APB, PDA, and TTX on

ERG responses recorded using both multifocal and

conventional methods in monkey: effects of APB,

PDA, and TTX on monkey ERG responses. Doc Oph-

thalmol 2002;105:189–222.

[13] Rangaswamy NV, Hood DC, Frishman LJ. Regional

variations in the local contributions to the primate

photopic flash ERG revealed using the slow-sequence

mfERG. Invest Ophthalmol Vis Sci 2003;44:3233–47.

[14] Hood DC, Greenstein VC, Holopigian K, et al. An

attempt to detect glaucomatous damage to the inner

retina with the multifocal ERG. Invest Ophthalmol

Vis Sci 2000;41:1570–9.

[15] Hood DC, Holopigian K, Seiple W, Greenstein V, Li J,

Sutter EE, et al. Assessment of local retinal function in

patients with retinitis pigmentosa using the multi-focal

ERG technique. Vision Res 1998;38:163–80.

[16] Greenstein VC, Holopigian K, Seiple W, Hood DC,

Carr RE. Non-detectable second-order responses in

patients with outer retinal diseases. Invest Ophthamol

Vis Sci 2001;42:384.

[17] Birch DG. The focal and multifocal electroretinogram.

In: Fishman GA, Birch DG, Holder GE, Brigell MG,

editors. Electrophysiologic testing in disorders of the

retina, optic nerve and visual pathway. San Francisco:

The Foundation of the American Academy of Ophthal-

mology 2001. p. 177–91.

[18] Hood DC, Zhang X. Multifocal ERG and VEP re-

sponses and visual fields: comparing disease-related

changes. Doc Ophthalmol 2000;100:115–37.

[19] Seeliger M, Kretschmann U, Apfelstedt-Sylla E,

Ruther K, Zrenner E. Multifocal electroretinography

in retinitis pigmentosa. Am J Ophthalmol 1998;125:

214–26.

[20] HolopigianK, SeipleW,GreensteinVC,HoodDC,Carr

RE. Local cone and rod system function in progres-

sive cone dystrophy. Invest Ophthalmol Vis Sci 2002;

43:2364–73.

[21] Hasegawa S, Ohshima A, Hayakawa Y, Takagi M, Abe

H. Multifocal electroretinograms in patients with

branch retinal artery occlusion. Invest Ophthalmol Vis

Sci 2001;42:298–304.

[22] Shimada Y, Horiguchi M. Stray light-induced multi-

focal electroretinograms. Invest Ophthalmol Vis Sci

2003;44:1245–51.

[23] Greenstein VC, Holopigian K, Hood DC, Seiple W,

Carr RE. The nature and extent of retinal dysfunction

associated with diabetic macular edema. Invest Oph-

thalmol Vis Sci 2000;41:3643–54.

[24] Fortune B, Schneck ME, Adams AJ. Multifocal elec-

troretinogram delays reveal local retinal dysfunction in

early diabetic retinopathy. Invest Ophthalmol Vis Sci

1999;40:2638–51.

[25] Sasoh M, Yoshida S, Kuze M, Uji Y. The multifocal

electroretinogram in retinal detachment. Doc Ophthal-

mol 1998;94:239–52.

[26] Si YJ, Kishi S, Aoyagi K. Assessment of macular func-

tion by multifocal electroretinogram before and after

macular hole surgery. Br J Ophthalmol 1999;83:420–4.

[27] Parks S, Keating D, Evans A, Williamson TH, Jay JL,

Ellis JA. Comparison of repeatability of the multifocal

electroretinogram and Humphrey perimeter. Doc Oph-

thalmol 1997;92:281–9.

[28] Nagatomo A, Nao-i N, Maruiwa F, Arai M, Sawada A.

Multifocal electroretinograms in normal subjects. Jpn J

Ophthalmol 1998;42:129–35.

[29] Verdon WA, Haegerstrom-Portnoy G. Topography of

the multifocal electroretinogram. Doc Ophthalmol

1998;95:73–90.

[30] Seiple W, Vajaranant TS, Szlyk JP, Clemens C,

Holopigian K, Paliga J, et al. Multifocal electroretino-

graphy as a function of age: the importance of norma-

tive values for older adults. Invest Ophthalmol Vis Sci

2003;44:1783–92.

[31] Hood D, Li J. A technique for measuring individual

multifocal ERG records: noninvasive assessment of

the visual system. Trends in Optics and Photonics

1997;11:33–41.

[32] Sutter EE, Bearse MA. The optic nerve head compo-

nent of the human ERG. Vision Res 1999;39:419–36.

[33] Hood DC, Bearse MA, Sutter EE, Frishman LJ. The

optic nerve head component of the monkey’s (Macaca

mulatta) multifocal electroretinogram (mERG). Vision

Res 2001;41:2029–41.

[34] Hood DC, Frishman LJ, Robson JG, Shady S, Ahmed J,

Viswanathan S. A frequency analysis of the regional

variation in the contribution from action potentials to

the primate multifocal ERG. In: Vision science and its

applications, Optical Society of America Tech. Dig.

Ser. Washington (DC): Opt Soc Amer; 1999. p. 56–9.

[35] Hood DC, Frishman LJ, Viswanathan S, Robson JG,

Ahmed J. Evidence for a ganglion cell contribution

to the primate electroretinogram (ERG): effects of TTX

on the multifocal ERG in macaque. Vis Neurosci

1999;16:411–6.

[36] Hare W, Ton H, Woldemussie E, Ruiz G, Feldmann B,

Wijono M. Electrophysiological and histological

measures of retinal injury in chronic ocular hypertensive

monkeys. Eur J Ophthalmol 1999;9(Suppl 1):30–3.

[37] Frishman LJ, Saszik S, Harwerth RS, Viswanathan S,

Li Y, Smith EL, et al. Effects of experimental glaucoma

in macaques on the multifocal ERG. Doc Ophthalmol

2000;100:231–51.

[38] Bearse MA, Sutter EE, Sim D, Stamper R. Glaucoma-

tous dysfunction revealed in higher order components

of the electroretinogram. In: Vision science and its ap-

plications, Optical Society of America Tech. Dig. Ser.

Washington (DC): Opt Soc Amer; 1996. p. 104–7.

[39] Hood DC, Greenstein V, Frishman L, Holopigian K,


Viswanathan S, Seiple W, et al. Identifying inner reti-

nal contributions to the human multifocal ERG. Vision

Res 1999;39:2285–91.

[40] Hasegawa S, Takagi M, Usui T, Takada R, Abe H.

Waveform changes of the first-order multifocal electro-

retinogram in patients with glaucoma. Invest Ophthal-

mol Vis Sci 2000;41:1597–603.

[41] Fortune B, Johnson CA, Cioffi GA. The topographic

relationship between multifocal electroretinographic

and behavioral perimetric measures of function in

glaucoma. Optom Vis Sci 2001;78:206–14.

[42] Sutter EE, Bearse MA, Stamper R. Monitoring retinal

ganglion cell function with the mfERG: recent ad-

vances. In: Vision science and its applications, Optical

Society of America Tech. Dig. Ser. Washington (DC):

Opt Soc Amer; 2001. p. 10–3.

[43] Palmowski AM, Allgayer R, Heinemann-Vemaleken

B. The multifocal ERG in open angle glaucoma—a

comparison of high and low contrast recordings in

high- and low-tension open angle glaucoma. Doc Oph-

thalmol 2000;101:35–49.

[44] Sutter EE, Bearse MA. The retinal topography of

local and lateral gain control mechanisms. In: Vision

science and its applications, Optical Society of Amer-

ica Tech. Dig. Ser. Washington (DC): Opt Soc Amer;

1998. p. 20–3.

[45] Bearse MA, Sutter EE, Stamper R. Detection of glau-

comatous dysfunction using a global flash multifocal

electroretinogram (mERG) paradigm. In: Vision sci-

ence and its applications, Optical Society of America

Tech. Dig. Ser. Washington (DC): Opt. Soc. Amer;

2001. p. 14–7.

[46] Shimada Y, Li Y, Bearse Jr MA, Sutter EE, Fung W.

Assessment of early retinal changes in diabetes using

a new multifocal ERG protocol. Br J Ophthalmol

2001;85:414–9.

[47] Fortune B, Bearse Jr MA, Cioffi GA, et al. Selective

loss of an oscillatory component from temporal retinal

multifocal ERG responses in glaucoma. Invest Oph-


[48] Palmowski AM, Allgayer R, Heinemann-Vernaleken

B, et al. Multifocal electroretinogram with a multiflash

stimulation in open-angle glaucoma. Ophthalmic Res

2002;34:83–9.

[49] Hood DC. Objective measurement of visual function

in glaucoma. Cur Opin Ophthalmol 2003;14:78–82.

[50] Baseler HA, Sutter EE, Klein SA, Carney T. The to-

pography of visual evoked response properties across

the visual field. Electroencephalogr Clin Neuro-

physiol 1994;90:65–81.

[51] Baseler HA, Sutter EE. M and P components of the

VEP and their visual field distribution. Vision Res

1997;37:675–90.

[52] Hood DC, Greenstein VC. The multifocal VEP and

ganglion cell damage: applications and limitations for

the study of glaucoma. Prog Retin Eye Res 2003;22:

201–51.

[53] Klistorner A, Graham SL. Objective perimetry in glau-

coma. Ophthalmology 2000;107:2283–99.

[54] Hood DC, Zhang X, Greenstein VC, et al. An inter-

ocular comparison of the multifocal VEP: a possible

technique for detecting local damage to the optic nerve.


[55] Fortune B, Hood DC. Conventional pattern-reversal

VEPs are not equivalent to summed multifocal VEPs.


[56] Zhang X, Hood DC. A principal component analy-

sis of the multifocal pattern reversal VEP. Journal of

Vision 2004, in press.

[57] Graham SL, Klistorner AI, Grigg JR, Billson FA. Ob-

jective VEP perimetry in glaucoma: asymmetry analy-

sis to identify early deficits. J Glaucoma 2000;9:10–9.

[58] Goldberg I, Graham SL, Klistorner AI. Multifocal ob-

jective perimetry in the detection of glaucomatous

field loss. Am J Ophthalmol 2002;133:29–39.

[59] Hood DC, Greenstein VC, Odel JG, et al. Visual field

defects and multifocal visual evoked potentials: evi-

dence of a linear relationship. Arch Ophthalmol 2002;

120:1672–81.

[60] Hood DC, Zhang X, Winn BJ. Detecting glaucoma-

tous damage with multifocal visual evoked potentials:

how can a monocular test work? J Glaucoma 2003;

12:3–15.

[61] Hood DC, Zhang X, Hong JE, Chen CS. Quantifying

the benefits of additional channels of multifocal VEP

recording. Doc Ophthalmol 2002;104:303–20.

[62] Zhang X, Hood DC, Chen CS, Hong JE. A signal-to-

noise analysis of multifocal VEP responses: an objec-

tive definition for poor records. Doc Ophthalmol 2002;

104:287–302.

[63] Hood DC, Odel JG, Zhang X. Tracking the recovery of

local optic nerve function after optic neuritis: a multi-

focal VEP study. Invest Ophthalmol Vis Sci 2000;41:

4032–8.

[64] Kardon RH, Givre SJ, Wall M, Hood D. Comparison

of threshold and multifocal-VEP perimetry in re-

covered optic neuritis. In: Wall M, Wills RP, editors.

Perimetry update 2000: proceedings of the XVII Inter-

national Perimetric Society Meeting. New York: Kug-

ler; 2001. p. 19–28.

[65] Hood D, Odel JG, Winn BJ. The multifocal visual

evoked potential (VEP): applications and limitations

in neuro-ophthalmology. J Neuroophthalmol, in press.

[66] Miele DL, Odel JG, Behrens MM, Zhang X, Hood D.

Functional bitemporal quadrantopia and the multifocal

visual evoked potential. J Neuroophthalmol 2000;20:

159–62.

[67] Quigley HA, Addicks EM, Green WR. Optic nerve

damage in human glaucoma. III. Quantitative correla-

tion of nerve fiber loss and visual field defect in glau-

coma, ischemic neuropathy, papilledema, and toxic

neuropathy. Arch Ophthalmol 1982;100:135–46.

[68] Quigley HA, Dunkelberger GR, Green WR. Retinal

ganglion cell atrophy correlated with automated

perimetry in human eyes with glaucoma. Am J Oph-

thalmol 1989;107:453–64.

[69] Quigley HA. Open-angle glaucoma. N Engl J Med

1993;328:1097–106.


[70] Klistorner AI, Graham SL, Grigg JR, Billson FA. Mul-

tifocal topographic visual evoked potential: improving

objective detection of local visual field defects. Invest

Ophthalmol Vis Sci 1998;39:937–50.

[71] Fortune B, Goh K, Demirel S, et al. Detection of glau-

comatous visual field loss using multifocal VEP.

Perimetry Update, in press.

[72] Hasegawa S, Abe H. Mapping of glaucomatous visual

field defects by multifocal VEPs. Invest Ophthalmol

Vis Sci 2001;42:3341–8.

[73] Hood DC, Thienprasiddhi P, Greenstein VC, Winn BJ,

Ohri N, Ritch R. Detecting early to mild glaucomatous

damage: a comparison of the multifocal VEP and au-

tomated perimetry. Invest Ophthamol Vis Sci 2004,

in press.


Multifocal visual evoked potential, multifocal

electroretinography, and optical coherence tomography in

the diagnosis of subclinical loss of vision

Thomas R. Hedges III, MD*, Maria-Luz Amaro Quireza, OD

Tufts University, New England Eye Center, 750 Washington Street, Box 450, Boston, MA 02111, USA

Neuro-ophthalmology has many possible defini- of the test. Although manual perimetric skills, once

tions, including the discipline of diagnosing diseases

affecting vision before the development of clearly

identifiable clinical signs. In addition, neuro-ophthal-

mology includes proving the absence of neurologic

disease, especially in patients with functional loss of

vision. Frequently, it is determined that a patient with

apparent neurogenic loss of vision actually has an

ophthalmic condition, such as occult maculopathy.

Three relatively new modalities are becoming techno-

logically refined enough to be useful in clinical prac-

tice. These methods are optical coherence tomography

(OCT), multifocal electroretinography (mfERG),

and multifocal visual evoked potential recording

(mfVEP). Perhaps the most revolutionary modality is

mfVEP, because it has the potential of replacing

conventional visual field testing.

Multifocal visual evoked potential

Perimetry remains one of the essential tools for

neuro-ophthalmic diagnosis; however, all forms of

visual field testing depend on the active participation

of the patient. The subjectivity of visual field testing

leads to considerable variability and unreliability. This

problem is especially true in automated perimetry in

which the examiner, and the interpreter of the visual

field, has less control over the patient’s performance


doi:10.1016/S0896-1549(03)00104-4



(T.R. Hedges III).

the pride of the neuro-ophthalmologist, are still neces-

sary, they are also time consuming. mfVEP record-

ing may provide a solution to these problems. If it can

be refined to the point at which small areas of visual

field loss can be detected, the ease of performance of

the test for the patient and the objective data provided

will advance the ability to diagnose conditions that

affect the visual field. Furthermore, the additional

physiologic information of latency and the amplitude

of the multifocal evoked potential waveform will add

more depth to the clinical information one needs to

make timely and accurate diagnoses.

For several decades, standard VEP has been used

to study a variety of ophthalmic problems [1]. This

technique essentially records macular responses. Mul-

tifocal technology [2] allows VEP waveforms to be

generated from stimuli to multiple areas of the visual

field [3,4]. This approach has been applied to a variety

of clinical disorders, particularly glaucoma [4,5]. The

authors have used the AccuMap system (ObjectiVi-

sion Pty, Sydney, Australia) and the Visual Evoked

Response Image System (VERIS; EDI, San Mateo,

California). This article discusses the authors’ prelim-

inary experience with both of these devices for neuro-

ophthalmic disorders.

The stimulus used for mfVEP includes families of

pseudorandom binary sequences of groups of 16

alternating checks (Fig. 1). These check patterns are

arranged within multiple segments of the stimulus

field (56 segments within 24 degrees [32 nasally] for

the AccuMap and 60 segments within 44.5 degrees for

the VERIS). Signals are recorded from two to four

electrodes arranged vertically or in a cross over the

occiput using single or multiple channels. The re-

s reserved.

Fig. 1. Multifocal VEP stimulus pattern (AccuMap).

T.R. Hedges III, M.L.A. Quireza / Ophthalmol Clin N Am 17 (2004) 89–10590

corded signals are amplified 100,000 times and

band pass filtered between 3 and 30 Hz. The signals

may be scaled to the electroencephalogram. Compu-

tation of signals occurs by cross-correlation of the

responses evoked by the sequence stimulation with

the sequences themselves. Refractive errors, cataracts,

and eccentric fixation can influence the results and

need to be considered when interpreting mfVEPs [6].

The authors have found that the results from

mfVEP correlate well with the results of automated

perimetry in many situations. Nevertheless, the results

from normal subjects are not entirely free of minor

variations and some variability. Interestingly, some of

the authors’ best normal recordings have been

obtained from patients who were presumed to have

functional or fictitious visual field defects (Fig. 2). In

fact, one of the best uses of mfVEP has been in

documenting normal visual field function in patients

with suspicious hemianopias and visual field constric-

tion. Identification of bitemporal hemianopia as a

functional problem on mfVEP has been reported by

Merle et al [7]. The authors have documented normal

responses in six patients, one with monocular hemi-

anopia, one with heteronymous quadrantanopias, one

with monocular severe visual field constriction, and

two with tubular visual fields (Massicott and Hedges,

submitted for publication, 2004).

The mfVEP results in patients who have macul-

opathy show findings that would be expected. The

amplitude and latency are clearly affected within the

area detected by perimetry (Fig. 3). In patients who

have long-standing visual field defects from optic

nerve and chiasmal disease, there seems to be a direct

correlation with automated visual field findings

(Fig. 4). More studies are needed to determine wheth-

er these apparent correlations will remain reliable. The

added feature of mfVEP of providing physiologic

data, as opposed to anatomic or simply localizing

information, is illustrated in patients with demyelin-

ating disease. As expected in patients with optic

neuritis, the mfVEP latencies are affected more pro-

foundly than are amplitudes. This effect can be seen in

patients who have recovered from optic neuritis, in

whom the amplitude of the mfVEP may be relatively

normal but in whom latency delays remain (Fig. 5).

Such latency delays may provide a more sensitive

means to follow the course of patients with optic

neuritis, especially when it occurs owing to multiple

sclerosis. Hood et al [8] reported persistent mfVEP

latency delays in three patients in whom loss of visual

acuity, visual field, and mfVEP amplitudes returned to

normal after treatment of optic neuritis. Patients with

ischemic optic neuropathy also have been studied with

mfVEP [9].

Multifocal VEP results from patients with retro-

chiasmal visual pathway disorders have not correlated

as well as expected with visual field findings. The

authors have found that mfVEP abnormalities do not

respect the vertical meridian as distinctly as visual

field changes (Fig. 6). This observation applies to

Fig. 2. Normal mfVEP responses (A) from a 45-year-old woman who had a temporal visual field defect in her left eye (B), which

remained when both eyes were tested simultaneously.

T.R. Hedges III, M.L.A. Quireza / Ophthalmol Clin N Am 17 (2004) 89–105 91

patients with long-standing occipital damage as well

as those with more recent damage to occipital and

visual radiations. Betsuin et al [10] found it necessary

to summate the responses in each quadrant of the

mfVEP to obtain useful data in patients with chiasmal

and retrochiasmal visual field defects. Wall et al [11]

found that 5 of 10 hemianopic visual field defects

were missed using a prototype of the AccuMap.

Multifocal VEP identifies visual field defects rela-

tively accurately and reproducibly. It allows an

analysis of latency and amplitude data at various

points in the visual field. A technician can perform

it rapidly and conveniently. Above all, it is objective.

Ultimately, mfVEP may replace visual field testing, at

least automated testing.

Multifocal electroretinography

Multifocal ERG, like mfVEP, permits an evalua-

tion of local electrophysiologic responses across the

human retina. It provides objective data useful for the

early detection of focal retinal disease and the evalua-

tion of treatment. The technique allows the simul-

taneous and topographical recording of the ERG

activity of hundreds of retinal locations from the pos-

terior pole, and it can separate true variation among

retinal regions from a variability in the responses over

time in a single recording session of 4 to 8 minutes. It

is especially useful in patients who have localized

damage in the outer retina. Conventional ERG re-

mains useful for detecting disorders that cause diffuse

Fig. 3. Multifocal VEP (A) from a patient with a well-circumscribed visual field defect (B) from a macular lesion.


retinal damage, such as retinitis pigmentosa and auto-

immune retinopathy.

The authors have been using the VERIS device

developed by Sutter et al [12] to record mfERGs. The

stimulus matrix consists of 103 hexagonal elements

within a 30-degree visual field at a viewing distance

of 27 cm. These elements are scaled in an array with

eccentricity to elicit more equal signal amplitudes at

all locations. The stimulation rate is 75 Hz. Each area

is stimulated with the same pseudorandom sequence

of light and dark, called a ‘‘maximum length se-

quence’’ or ‘‘m-sequence,’’ so that each video frame

has a 50% probability of being light or dark. In this

way, the overall luminance remains constant. Because

of the light levels employed and the repetitive stim-

ulation, cone responses are recorded. Burian-Allen

bipolar contact lens electrodes are used. The subject’s

pupils are dilated. The opposite eye is occluded. After

visual acuity is corrected for the viewing distance,

patients are instructed to maintain fixation on a

central spot. This fixation can be monitored with a

camera. The first-order components of the mfERG

cone responses are recorded, and three-dimensional

display plots are obtained in each case. An artifact

Fig. 4. Multifocal VEP (A,B) from a man with bitemporal visual field defects (C,D) from a pituitary adenoma.


Fig. 5. Multifocal VEPs from a 30 year-old woman who had recovered most of the vision lost from optic neuritis associated with

multiple sclerosis. mfVEP for the left eye is normal (A) but shows increased latencies in the right mfVEP (B). Visual acuities

were 20/15 OD and 20/20 OS, and visual fields were relatively normal OS (C) and showed areas of mild depression OD (D).


elimination technique is used. The first-order kernel

can be considered as the mean response to a local

flash. It consists of a negative deflection followed by

a positive deflection, with a shape similar to that of

conventional ERG a- and b-waves. Whether these

deflections have the same retinal origins as ERG

a- and b-waves remains to be determined [13]. The

authors have found mfERG to be useful in diagnosing

a variety of relatively occult macular disorders, in-

cluding hydroxychloroquine retinopathy, Stargardt’s

disease, acute idiopathic blind spot enlargement syn-

drome, and occult age-related macular degeneration.

Retinopathy from hydroxychloroquine remains a

difficult problem to diagnose early. No broadly ac-

cepted screening tests exist. Full-field ERG is incapa-

ble of detecting early damage, which appears to be

located in the macular area. Maturi et al [14] described

a case in which a 45-year-old woman was treated

with hydroxychloroquine averaging 400 mg/day

for 6 years. mfERG showed decreased electrical re-

sponses in the macular areas with a reduction in

amplitude and latency. Visual symptoms and loss of

visual acuity had already become well established

[14]. Recently, the authors reported a series of cases

in which mfERGs were distinctly abnormal in three

patients in whom visual acuities were still 20/20 to

20/25 (Fig. 7) [15]. These patients were symptomatic

and had detectable perifoveal visual field loss. mfERG

confirmed that the visual symptoms were caused by

hydroxychloroquine. To date, the authors have not

detected mfERG abnormalities in patients taking

hydroxychloroquine who have not become sympto-

matic. Nevertheless, identifying presymptomatic

mfERG abnormalities in these patients through pro-

Fig. 6. Multifocal VEP (A,B) from a 55-year-old woman who had right homonymous hemianopia (C) from apparent perinatal

hypoxic injury to the left occipital cortex since birth associated with hemiband atrophy of the optic nerves.


spective follow-up at various intervals of time might

allow for discontinuation of the drug before visual

morbidity occurs.

The acute idiopathic blind spot enlargement syn-

drome and the multiple evanescent white dot syn-

drome, which may be distinctive phenomena, can

usually be diagnosed with confidence by clinical

examination. In some cases, confirmation that blind

spot enlargement is caused by peripapillary retinal

disease may be necessary to avoid more expensive or

invasive testing to rule out other disorders. mfERG

has been reported to show distinctive peripapillary

changes in patients who have acute idiopathic blind

spot enlargement syndrome or the multiple evanescent

white dot syndrome [16–20]. The authors also have

found mfERG to be helpful in such cases (Fig. 8).

Stargardt’s disease frequently presents as a loss of

visual acuity in children who have minimal findings

on clinical examination. If flecks are not seen in the

retina, one must rely on fluorescein angiographic

absence of normal choroidal fluorescence, the so-

called ‘‘silent choroid.’’ Scanning laser ophthalmos-

copy can show macular changes that can identify

subgroups of Stargardt’s retinopathy [21]. Kretsch-

mann et al [22] found that, in patients with Stargardt’s

retinopathy, mfERG could detect foveal dysfunction

Fig. 7. Multifocal ERG recordings (A) showing generalized depression surrounding a relatively preserved central peak from a

52-year-old woman who had been on hydroxychloroquine for 8 years. She had visual acuities of 20/25 OD and 20/25 OS, normal

screening color plate testing, and pericentral visual field loss on automated perimetry (B) compared with a normal mfERG (C).


Fig. 8. Multifocal ERG showing reduced responses in the area corresponding with blind spot enlargement (A) from a 38-year-old

woman. (B) Visual acuity was 20/20 in both eyes, and there was mild haziness around an otherwise normal-appearing optic disk

and retina.


in the early stages of the disease in patients lacking

morphologic fundus changes and with good visual

acuity. The authors also have found mfERG to be

useful in diagnosing Stargardt’s disease (Fig. 9).

Multifocal ERG is also useful in occult maculo-

pathy from age-related macular degeneration [23],

retinitis pigmentosa [24], and branch retinal artery

occlusion [25]. Piao et al reported eight cases of occult

maculopathy in which the macular ERG amplitudes

and implicit times were decreased and delayed, re-

spectively, when ophthalmoscopic, fluorescein angi-

ographic, and full-field ERG results were normal.

Chappelow and Marmor [26] recorded the mfERG

in six patients who had central serous choroidopathy.

The mfERG was depressed not only in the areas of

detachment but also beyond the detachment in the

acute phase of the disorder. In five patients, after

resolution of clinical signs, mfERG amplitudes im-

proved but remained subnormal throughout the pos-

terior pole, and the latencies improved from prolonged

to midnormal values [26]. Vajaranant et al [27] found

mfERG abnormalities in clinically affected regions

only in six patients. Other conditions in which mfERG

has been employed include areolar choroidal dystro-

phy [28], Best’s macular dystrophy [29], and macular

holes [30]. Although mfERG findings have been

recorded in patients with glaucoma [31,32], the clini-

cal usefulness of these findings in the management of

glaucoma remains to be seen.

Multifocal ERG can be used to rule out retinal

disease. When one wishes to confirm normal central

retinal function to bolster the evidence for the location

of a lesion responsible for visual loss elsewhere in the

visual system, such as a subclinical cataract, a normal

mfERG can be helpful. The mfERG should be normal

in most optic nerve diseases; however, caution must

be maintained in cases in which visual acuity is

reduced to the point where fixation may not be

adequate throughout the mfERG test or, if fixation is

maintained, it is eccentric. The authors have found

that, in patients with acute Leber hereditary optic

neuropathy, initial mfERGs can appear abnormal. In

Fig. 9. Multifocal ERGs (A,B) from a 13-year-old boy with visual acuity of 20/60 OD and 20/40 OS. There were no fundus

abnormalities except for a dark choroid on fundus fluorescein angiography (C,D). Full-field ERG was normal.


Fig. 10. Optical coherence tomogram (A) and fluorescein angiogram (B) from a 34-year-old man with a history of facial

numbness and blotchy vision in the right eye for 1 week. Visual acuity was 20/15 OD and 20/10 OS, color vision normal, there

was no afferent pupillary defect, and visual field testing was unrevealing. OCT showed a discrete area of subretinal fluid. Review

of the fluorescein angiogram showed a subtle area of leakage in the macula.



patients with long-standing visual loss from Leber’s

neuropathy, mfERGs are better and practically normal

in some patients. The apparent abnormalities on

mfERG in acutely affected patients most likely rep-

resent an artifact of fixation. These abnormalities have

been noted by other investigators (R.H. Karden,

personal communication, 2002).

Multifocal ERG is an efficient clinical technique to

identify and map abnormalities in the outer retina in

the early phases of diseases not detectable with other

methods or in cases in which ophthalmic findings are

sparse or inconclusive. mfERG may help distinguish

different retinal diseases and localize the site of the

dysfunction. Proving the presence of retinal dysfunc-

tion by mfERG can allow clinicians to avoid expen-

sive and worrisome neurologic investigations.

Optical coherence tomography

Optical coherence tomography is another clinical

tool that can be helpful in neuro-ophthalmic situa-

tions. It is very sensitive for identifying and docu-

Fig. 11. A 9-year-old boy with neuroblastoma was treated with o

within 1 week. MRI was normal. (A, B) OCT showed subretinal an

tinct margins.

menting subtle abnormalities within and under the

retina in patients thought to have possible optic

neuropathy because of a paucity of clinical and

angiographic retinal findings. OCT can identify nerve

fiber layer loss that may be difficult to document

photographically. OCT can be used to identify, docu-

ment, and measure retinal nerve fiber layer thickening

from papilledema and conditions in which there is

swelling of the optic nerve and the nerve fiber layer.

OCT has been useful in demonstrating secondary

retinal effects from optic neuropathy. It adds to the

repertoire of techniques that provide objective and,

most importantly, quantitative data useful in neuro-

ophthalmic diagnosis and treatment. The technical

aspects of OCT have been well described elsewhere

[33]. The following sections describe some examples

of neuro-ophthalmic uses of OCT.

Retinopathy suggesting optic neuropathy

Frequently, neuro-ophthalmologists are asked to

see patients for whom the clinical retinal findings do

not appear to account for the degree of visual loss

ccipital radiation and cisplatin and lost vision in both eyes

d intraretinal edema. (C, D) Both optic disks showed indis-


detected by visual acuity or visual field testing. Many

of these patients are referred by retinal surgeons who

have problems dealing with retinal a condition for

which there is no treatment. If they cannot repair the

problem, they have trouble accepting that it accounts

for the patient’s clinical syndrome. Proving that a

retinal lesion is indeed responsible for a given visual

defect is all too often the duty of the neuro-ophthal-

mologist. Frequently, this duty leads to making a

diagnosis by exclusion. Such an approach can become

expensive and time consuming.

Some patients with age-related macular degen-

eration may have minimal abnormalities that can

be visualized ophthalmoscopically or on fluorescein

angiography. OCT has been shown to be helpful in

demonstrating retinal and subretinal abnormalities,

proving that the location of an apparently subclinical

loss of vision is retinal in origin [33]. In almost all

situations, central serous retinopathy can be diagnosed

Fig. 12. A 30-year-old nurse experienced memory loss and fatigue

acuities were 20/20 in both eyes, but there were persistent left h

thinning of the nerve fiber layer in the contralateral right eye and l

by ophthalmoscopy. Usually, it is easily confirmed by

fluorescein angiography; however, in rare cases, OCT

can show small collections of subretinal fluid that may

not be clearly seen otherwise. OCT can readily show

subretinal fluid even in the most mild cases (Fig. 10).

OCT scans can be performed quickly through an

undilated pupil, which can be important in patients

suspected of having optic neuritis in whom afferent

pupillary function may need to be monitored.

Cisplatin has been reported to cause optic neurop-

athy, cerebral leukomalacia, and retinopathy [34]. The

authors found that OCT revealed retinal effects of

cisplatin in one patient (Fig. 11).

Nerve fiber layer loss

Most of the work on OCT measurement of nerve

fiber layer changes from optic atrophy has been in the

. MRI showed a craniopharyngioma. Postoperatively, visual

omonymous visual field defects (A). OCT showed diffuse

oss temporally (B) and nasally (C) in the ipsilateral left eye.

Fig. 13. A 70-year-old woman had anterior ischemic optic neuropathy in her right eye with optic disk swelling and inferior visual

field loss. The left visual field (A) and acuity were normal. OCT showed nerve fiber layer thickening in the left asymptomatic eye

(B), and re-evaluation of the left optic nerve head showed mild swelling and a peripapillary flame hemorrhage (C). Three weeks

later, she had anterior ischemic optic neuropathy in the left eye with inferior visual field loss (D), more severe nerve fiber layer

swelling (E,F).


T.R. Hedges III, M.L.A. Quireza / Ophtha

field of glaucoma [33]. In patients who have neuro-

ophthalmic problems, OCT demonstration of retinal

nerve fiber layer thinning can be used to detect and

monitor optic atrophy from a variety of conditions,

including optic nerve head drusen [35], optic pits [36],

optic nerve trauma [37], and optic hypoplasia [38].

OCT also can be used to help rule out optic neurop-

athy when nerve fiber layer measurements are normal.

Optical coherence tomography can be used to

measure retinal nerve fiber layer loss in conditions

affecting the pregeniculate visual pathways and in rare

cases when retrogeniculate lesions have been present

long enough to cause transsynaptic degeneration.

The utility of OCT nerve fiber layer analysis for pa-

tients with lesions such as pituitary adenoma remains

to be determined. Currently, visual field testing (and

mfVEP) to detect optic nerve dysfunction before

nerve fiber layer loss occurs remains the most effec-

tive way to assist endocrinology and neurosurgical

colleagues in following patients with compressive

optic neuropathy (Fig. 12).

Fig. 14. Fundus photographs (A,B) and OCT showing submacu

intracranial hypertension. Visual acuities were 20/50 OD and 20/6

Retinal nerve fiber layer thickening

Optical coherence tomography can demonstrate

and measure retinal nerve fiber layer thickening. It

may be useful in detecting subclinical anterior ische-

mic optic neuropathy. The authors have seen OCT

abnormalities in patients who presented with anterior

ischemic optic neuropathy in one eye weeks before

they had clinical signs and symptoms in the opposite

eye, a presentation that has been observed clinically

by others (Fig. 13) [39].

Optical coherence tomography is useful in detect-

ing and following pathologic optic disk swelling in

patients with papilledema. During the course of the

authors’ investigations regarding the usefulness of

OCT in the diagnosis and management of papille-

dema, macular scans in some patients revealed sub-

retinal abnormalities. These findings occurred in

patients who had a loss of visual acuity, which can

also occur from intraretinal macular edema (macular

star) and macular hemorrhage. Histopathologic evi-

lmol Clin N Am 17 (2004) 89–105 103

lar fluid (C,D) from a 30-year-old woman with idiopathic

0 OS.


dence of subretinal fluid associated with papilledema

has been reported [40], but clinical evidence has been

limited [41]. In a series of 45 patients who had

bilateral optic disk edema from increased intracranial

pressure, the authors detected subretinal fluid on OCT

that resolved after treatment in six cases (Fig. 14) [42].

Summary

Multifocal VEP and mfERG techniques show

promise for routine clinical use in neuro-ophthalmol-

ogy. OCT is now used in many subspecialty areas of

ophthalmology, especially in studies of the retina and

glaucoma. Although neuro-ophthalmologists must

still rely on time-consuming history taking and careful

clinical examination, the three modalities described

herein can provide objective proof regarding clinical

diagnoses in selected cases, saving time and avoiding

costly neurologic investigations.

References

[1] Sokol S. Visually evoked potentials: theory, technique

and clinical application. Surv Ophthalmol 1976;21:

18–44.

[2] Sutter EE. The fast m-transform: a fast computation of

cross correlations with binary m-sequences. Soc Ind

Appl Math 1991;26:686–94.

[3] Baeseler HA, Sutter EE, Klein SA, Carney T. The to-

pography of visual evoked response properties across

the visual field. Electroencephalogr Clin Neurophysiol

1994;90:65–81.

[4] Hood DC, Greenstein VC. Multifocal VEP and gan-

glion cell damage: applications and limitations for

the study of glaucoma. Prog Retin Eye Res 2003;22:

201–51.

[5] Graham SL, Klistorner AI, Grigg JR, Billison FA. Ob-

jective perimetry in glaucoma: asymmetric analysis to

identify early deficits. J Glaucoma 2000;9:10–9.

[6] Winn BJ, Hood DC, Shin EH, Odel JG. Interpreting the

multifocal visual evoked potential (mfVEP): the influ-

ence of preretinal factors. Presented at the North Ameri-

can Neuro-ophthalmology Society Annual Meeting.

Salt Lake City, April, 2003.

[7] Merle DL, Odel JG, Berhens M, Zhang X, Hood DC.

Functional bitemporal quadrantanopia and the multifo-

cal visual evoked potential. J Neuroophthalmol 2000;

20:159–62.

[8] Hood DC, Odel JG, Zhang X. Tracking the recovery of

local optic nerve function after optic neuritis: a multi-

focal VEP study. Invest Ophthalmol Vis Sci 2000;41:

4032–8.

[9] Hood DC, Zhand X, Greenswtein VC, Kangovi S,

Odel JG, Liebmann JM, et al. An interocular compari-

son of the multifocal VEP: a possible technique for

detecting local damage to the optic nerve. Invest Oph-


[10] Betsuin Y, Mashima Y, Ohde H, Inoue R, Oguchi Y.

Clinical application of the multifocal VEP’s. Curr Eye

Res 2001;22:54–63.

[11] Wall M, Woodward K. The multifocal visual evoked

response in neuro-ophthalmic diagnosis. Presented at

the North American Neuro-ophthalmology Society An-

nual Meeting. Salt Lake City, April, 2003.

[12] Sutter EE, Tran D. The field topography of ERG com-

ponents in man. I. The photopic luminance response.

Vision Res 1992;32:433–66.

[13] Hood DC, Frishman LJ, Saszik S. Retinal origins of

the primate multifocal ERG: implications for the hu-

man response. Invest Ophthalmol Vis Sci 2002;43:

1676–85.

[14] Maturi RK, Folk JC, Nichols B, Oetting TT, Kardon

RH. Hydroxychloroquine retinopathy. Arch Ophthal-

mol 1999;177:1262–3.

[15] So SC, Hedges TR, Schuman JS, Amaro MLQ. Evalu-

ation of hydroxychloroquine retinopathy with multifo-

cal electroretinography. Ophthalmic Surg Lasers 2003;

34:251–8.

[16] Oh KT, Folk JC, Maturi RK, Kardon RH. Multifocal

electroretinography in multifocal choroiditis and the

multiple evanescent white dot syndrome. Retina 2001;

21:581–9.

[17] Huang HJ, Yamazaki H, Kawabata H, Ninomiya T,

Adachi-Usami E. Multifocal electroretinogram in mul-

tiple evanescent white dot syndrome. Doc Ophthalmol

1996;92:301–9.

[18] Arai M, Nao-i N, Sawada A, Hayashida T. Multifocal

electroretinogram indicates visual field loss in acute

zonal occult outer retinopathy. Am J Ophthalmol

1998;21:126:466–9.

[19] Volpe NJ, Rizzo JF, Lessell S. Acute idiopathic blind

spot enlargement syndrome: a review of 27 new cases.


[20] Watzke RC, Shults WT. Clinical features and natural

history of the acute idiopathic enlarged blind spot syn-

drome. Ophthalmology 2002;109:1326–35.

[21] Lois N, Holder GE, Bunce C, Fitzke FW, Bird AC.

Phenotypic subtypes of Stargardt macular dystrophy—

fundus flavimaculatus. Arch Ophthalmol 2001;119:

359–69.

[22] Kretschmann U, Seeliger MW, Ruether K, Usui T,

Apfelstedt-Sylla E, Zrenner E. Multifocal electroreti-

nography in patients with Stargardt’s macular dystro-

phy. Br J Ophthalmol 1998;82:267–75.

[23] Piao CH, Kondo M, Tanikawa A, Terasaki H, Miyake

Y. Multifocal electroretinogram in occult macular dys-

trophy. Invest Ophthalmol Vis Sci 2000;141:513–7.

[24] Hood DC, Holopigian K, Seiple W. Assessment of

local retinal function in patients with retinitis pigmen-

tosa using the multifocal electroretinogram. Vision Res

1998;38:163–80.

[25] Hasegtawa S, Ohshima A, Hayakawa Y. Multifocal

electroretinograms in patients with branch retinal ar-


tery occlusion. Invest Ophthalmol Vis Sci 2002;43:

298–304.

[26] Chappelow AV, Marmor MF. Multifocal electroretino-

gram abnormalities persist following resolution of cen-

tral serous chorioretinopathy. Arch Ophthalmol 2000;

118:1211–5.

[27] Vajaranant TS, Szlyk JP, Fishman GA, Gieser JP, Seiple

W. Localized retinal dysfunction in central serous cho-

rioretinopathy as measured using the multifocal electro-

retinogram. Ophthalmology 2002;109:1243–50.

[28] Nagasaka K,HoriguchiM, ShimadaY, YusawaM.Mul-

tifocal electroretinograms in cases of central areolar

choroidal dystrophy. Invest Ophthalmol Vis Sci 2003;

44:1673–9.

[29] Scholl HP, Schuster AM, Vonthien R, Zrenner E. Map-

ping of retinal function in Best macular dystrophy using

multifocal electroretinography. Vision Res 2002;42:

1053–61.

[30] Oh KT, Boldt C, Maturi RK, Folk JC, Kardon RH.

Evaluation of patients with visual field defects follow-

ing macular hole surgery using multifocal electroreti-

nography. Retina 2000;21:238–43.

[31] Hasegawa S, Takagi M, Usui T. Waveform changes of

the first-order multifocal electroretinogram in patients

with glaucoma. Invest Ophthalmol Vis Sci 2000;41:

1597–603.

[32] Hood DC, Greenstein VC, Holopigian K. An attempt to

detect glaucomatous damage to the inner retina with the

multifocal ERG. Invest Ophthalmol Vis Sci 2000;41:

1570–9.

[33] Schuman JS, Puliafito C, Fujimoto JG, editors. OCT

atlas of ocular diseases. 2nd edition. Thorofare (NJ):

Slack; in press.

[34] Katz BJ, Ward JH, Digre KB, Creel DJ, Mamalis N.

Persistent severe visual loss and electroretinographic

abnormalities after intravenous cisplatin therapy.

J Neuroophthalmol 2003;23:132–5.

[35] Roh S, Noecker RJ, Schuman JS, Hedges TR, Weiter

JJ, Mattox C. Effect of optic nerve head drusen on

nerve fiber layer thickness. Opthalmology 1998;105:

878–85.

[36] Theodossiadis GP, Theodossiadis P. Optical coherence

tomography in optic disc pit maculopathy treated by

macular buckling procedure. Am J Ophthalmol 2001;

132:184–90.

[37] Medeiros FA, Moura FC, Vessani RM, Susanna R.

Axonal loss after traumatic optic neuropathy docu-

mented by optical coherence tomography. Am J Oph-

thalmol 2001;135:406–8.

[38] Unoki K, Ohba N, Hoyt WF. Optical coherence tomog-

raphy of superior segmental optic hypoplasia. Br J

Ophthalmol 2002;86:910–4.

[39] Hayreh SS. Anterior ischemic optic neuropathy. V. Op-

tic disc edema an early sign. Arch Ophthalmol 1981;99:

1030–40.

[40] Samuels B. Histopathology of papilledema. Am J Oph-

thalmol 1938;21:1242–8.

[41] Corbett JJ, Jacobson MD, Mauer MD, Thompson HS.

Enlargement of the blind spot caused by papilledema.

Am J Ophthalmol 1998;105:261–5.

[42] HoyeVJ, Berrocal AN, Hedges TR, Amaro-Qui-

reza ML. Optical coherence tomography demon-

strates subretinal macular edema from papilledema.


Imaging

Documents

Transcript of Imaging