The use of magnetic resonance imaging in thediagnosis of neurological disease

Wendy A. Stewart, Joane M-L. Parent, Rheal A. Towner, Howard Dobson

AbstractMagnetic resonance imaging (MRI) data were cor-related with clinical and cerebrospinal fluid (CSF) find-ings in one cat and two dogs with brain lesions. In allthree cases, localization of the lesions, as determinedclinically, was confirmed using MRI. Magneticresonance imaging also helped us to define the fullextent of the lesion(s) in each case. In one case, thelesion would have been diagnosed. as purely inflam-matory based on the abnormalities in the CSF. TheMRI study, however, showed a homogeneous masswith circumferential changes characteristic ofperitumoral edema or inflammation. In two cases, theMRI findings were confirmed at necropsy. An MRIstudy was also done on a normal dog, demonstratingthe variable contrast and anatomical detail possibleusing this technique. We also discuss difficulties inidentifying tumor type using MRI.

R6sumeL'utilisation de lilmagerie par la r6sonnancemagn6tique pour le diagnostic de maladlesneurologilquesLes donnees obtenues lors d'examen a l'aide del'imagerie par resonnance magnetique (I.R.M.) ont etecorrelees avec les signes cliniques et les analyses duliquide cephalo-rachidien chez un chat et deux chienspresentant des lesions cerebrales. Dans les trois cas,la localisation de la lesion, determinee par l'examenphysique, a ete confirmee par l'I.R.M. Ce moyen dediagnostic a aussi aide A determiner l'etendue de la oudes lesions dans chacun des cas. Dans un cas, la lesionaurait ete diagnostiquee comme etant inflammatoireselon les resultats de l'analyse du liquide cephalo-rachidien. Toutefois, I'utilisation de l'imagerie parresonnance magnetique a permis de deceler une massehomogene presentant un oedeme ou une inflammationcirconferentielle peritumorale. Dans deux cas, lesresultats de l'I.R.M. ont ete confirmes lors d'uneautopsie. Des examens ont et6 effectues chez un chienen sante afin de demontrer les caracteristiques

MRI Facility, Room 1601, Biomedical Sciences Building(Stewart, Towner) and Department of Clinical Studies(Parent, Dobson), Ontario Veterinary College, Universityof Guelph, Guelph, Ontario NIG 2W1.

Can Vet J Volume 33, September 1992

anatomiques et les variations de contraste de la region,lors de l'utilisation de cette technique de diagnostic.Les auteurs discutent aussi des difficultis assocides Al'identification du type de tumeur par I.R.M.

(Traduit par Dr Th6rase Lanthier)

Can Vet J 1992; 33: 585-590

IntroductionAlthough magnetic resonance imaging (MRI) isAroutinely used in human medicine to diagnosebrain disease (1,2), its use in veterinary medicine isinfrequent. This is due to: 1) the high purchase andmaintenance costs of the equipment, 2) the limitedavailability of, and access to, preexisting facilities,and 3) the fact that its potential is unknown to veteri-narians. The MRI Facility at the Ontario VeterinaryCollege (OVC) is one of only two presently locatedwithin veterinary institutions in North America.

In a number of veterinary institutions, X-raycomputed tomography (CT) is used to aid cliniciansin the diagnosis of neurological disease. Computedtomography and MRI are similar in that they providecross-sectional images of the patient, based on anumerically derived, computerized representation ofcertain physical properties of tissue. The contrastobserved in CT images is dependent on tissue and elec-tron density. Magnetic resonance imaging usesmagnetic fields to obtain images of the soft tissues ofthe body. It is noninvasive and in most studies doesnot require the use of a contrast agent to visualizeabnormalities. In addition, because no ionizing radia-tion is used, it allows repeated study of the samepatient over time. It is also possible to obtain imagesin three orientations (transverse, sagittal, and coronal)without repositioning the patient. There are severalcontraindications to the use of MRI. One is the pres-ence of ferromagnetic materials in the body, such asintracranial aneurysm clips, as the magnetic field maycause them to move. Cardiac pacemakers may mal-function in the magnetic field. The use of gasanesthetic machines containing ferromagnetic mate-

585

rials is not a problem with small bore magnets,provided that the anesthetic circuit is extended to pro-vide a safe distance from the magnet. The distancerequired is dependent on the bore size and the magneticfield strength. For veterinary cases with increased CSFpressure, cardiovascular disease, or renal disease, thescanning time (1-2 hours) may also be a risk.A previous study using four dogs correlated MRI

data with normal anatomical features of the caninebrain (3). Our intent here is to introduce veterinaryclinicians to the capabilities of MRI in aiding diagnosisof brain disease in dogs and cats. This is achieved usingthe scan of a normal brain showing gross anatomicalstructures and the variable contrast possible usingMRI, followed by three case reports where MRIenhanced the clinical diagnosis.

Materials and methodsMagnetic resonance imaging obtains images of themobile water and lipid protons within tissues. In nor-mal brain, lipid protons are bound in the complex,structural lipoproteins comprising myelin. Theresulting restricted motion of the lipid protons givesrise to a short spin-spin relaxation time (T2), whichcauses the signal from these protons to decay veryrapidly and contribute little or no signal intensity tothe normal brain image. Diseased states which affectmembrane structure give rise to mobile lipid protonswhich can contribute to the signal observed on amagnetic resonance image. Protons can be observedusing this technique due to a property of their nucleiknown as spin. This property causes the nuclei toexhibit certain characteristics in the presence of amagnetic field: most protons align themselves with(parallel to) the direction of the magnetic field and theremainder are aligned against (antiparallel to) the field.The MRI instrument consists of a magnet with a

specified magnetic field strength. The homogeneity ofthe magnetic field varies along the magnet's length:diagnostic quality images are obtained when the regionof interest in the patient is placed in the mosthomogeneous region of the magnetic field, which isusually close to the centre of the magnet. In additionto the static magnetic field produced by the magnet,a second, variable field, perpendicular to the first, isgenerated by a radiofrequency (rf) coil surrounding theregion of interest in the patient (Figure 1). This secondfield is applied as pulses of a predetermined length andamplitude, to affect the proton spins in a specific man-ner. Combinations of these pulses are referred to aspulse sequences. A typical pulse sequence is shown inFigure 2. The timing in these pulse sequences can bemanipulated to provide variable contrast in the finalimage. This is possible due to differences in themagnetic properties of water protons in differenttissues. After application of a pulse sequence, the pro-tons return to their equilibrium state via processesknown as relaxation. Two time constants are asso-ciated with this return to equilibrium: the spin-latticerelaxation time (T1) and the spin-spin relaxation time(T2). For reasons which are not fully understood,there are inherent differences in tissue T, and T2values which provide a naturally occurring contrast in

Figure 1. Imaging set-up for veterinary cases, showing adog positioned on the cradle with its head inside the radio-frequency coil. The complete set-up is pushed inside the boreof the magnet. A. Superconducting magnet. B. Radio-frequency coil. C. Cradle used to hold the animal.

900 1800

dl 1 TE/2 4 lTE/2

RF \ V

Signal A

Echo

Figure 2. Spin-echo pulse sequence. dl = delay time andTE = time of the echo or signal. TR = (dl + TE + halfthe 900 pulse length + acquisition time) and is called therepeat time or total time before reapplication of the entirepulse sequence.

the image. In addition, these parameters are sensitiveto the chemical environment; therefore any patho-logical changes which alter the tissue chemical envi-ronment, including water content, give rise to changesin T1 and T2. It is possible to "weight" theappearance of the final image towards one parameter;that is, the contrast observed is mainly due to dif-ferences in either proton density, T1 or T2. A shortecho time (TE) and repeat time (TR) (see Figure 2) isreferred to as a TI-weighted image, and a long TEand TR, a T2-weighted image. It is also possible toobtain images that are dependent on both T, and T2where TE is long and TR is short. Lesions can bevisualized on the MRI due to changes in Ti and/orT2 in the abnormal area or due to gross structuralchanges. T1-weighted images provide exquisiteanatomical detail, but do not always provide optimalcontrast between normal and abnormal brain tissue.The choice of pulse sequence is pathology-dependent.For example, if inflammation is thought to be present,a T2-weighted image would be the preferred choice.However, to demonstrate a herniated invertebral disc,a TI-weighted image is the optimal choice.

Magnetic resonance imaging data were collected ona SISCO 85/310 (Spectroscopy Imaging Systems,Fremont, California, USA) animal imaging systemdesigned for experimental animal studies. It is com-

586 Can Vet J Volume 33, September586 Can Vet J Volume 33, September 1992

prised of a superconducting magnet with a 31 cm bore,operating at a magnetic field strength of 2.0 Tesla.This corresponds to a proton resonance frequency of85 MHz. A 13 cm diameter rf coil (see Figure 1) wasused to acquire the data. Images were obtained in thetransverse orientation. Multiple slices, 2.3 mm thick,were obtained through the brain using T1-(TE =25 ms, TR = 1 sec) and T2-weighted (TE = 100 ms,TR = 3-4 sec) and combination weighted (TE =60 ms, TR = 1 sec) pulse sequences. These pulsesequences were chosen to provide the anatomical detailrequired to localize the lesion and to determine itsextent in the brain.

SubjectsOne normal dog and three neurological patients (twodogs and one cat) were maintained under anesthesiausing halothane. Anesthesia is necessary to keep theanimals still throughout the scanning procedure. Theywere positioned in ventral or dorsal recombency insidethe rf coil and placed inside the magnet (see Figure 1).Imaging data were collected as described above. Fol-lowing imaging, the normal dog was euthanized,perfusion-fixed, and frozen. The brain was then sec-tioned to correlate with the acquired MRI data. In twoof the clinical cases, the animals were humanelydestroyed and their brains removed and fixed in 10%neutral buffered formalin. They were subsequently sec-tioned, and representative samples taken for histolog-ical examination. The third case was diagnosed as aninflammatory process, which responded to medicaltreatment.

ResultsNormal AnatomyFigures 3A and B show magnetic resonance images,and Figure 3C, the corresponding gross section fromthe normal dog's brain. On T1-weighted images(Figure 3A), the gross anatomy is well defined, butcontrast within the brain parenchyma is limited andthe CSF appears dark. On the T2-weighted images(Figure 3B), the architecture of the brain is clearlydefined and CSF is white. The CSF in the sulci is alsodefined, and must not be mistaken for lesions such asinflammation or, in cases with large sulci, a tumor.The contrast shown in these images is typical of anymammalian brain. We have noted that the canine andfeline brains have differences in shape, ventricle size,and size and shape of sulci, but the grey/white mattercontrast and CSF signal are virtually identical.

Case IHistory and localization of lesionA sixteen-year-old female, spayed, domestic longhaired cat was presented to OVC with a two-year his-tory of episodic collapses. The episodes (seizure/behavioral changes) were described as lasting only afew minutes during which the animal was unaware ofits surroundings, uncoordinated, and fell on its side.Following recovery, it sometimes circled to the right.The episodes were increasing in frequency. For the lastfour months, the animal resented being brushed on theleft side of the body. Physical examination revealed

Can Vet J Volume 33, September 1992

Figure 3. A. TI-weighted MRI from normal dog brain.TE = 25 ms, TR = 1 sec. B. T2-weighted MRI from sameslice as in A. TE = 100 ms, TR = 3 sec. 1. grey matter,2. white matter, 3. lateral ventricles, 4. sulcus. C. Posteriorsurface of 5 mm thick slice at the equivalent level.

587

Figure 4. Combined Ti- and T2-weighted image throughthe parietal lobe in case 1. TE = 60 ms, TR = 1 sec.Arrows show extent of meningioma.

no abnormalities. The neurological abnormalities werelimited to marked hyperesthesia over the whole leftside of the body and head. There was no ataxia butthe animal appeared lame in the left fore and hindlimbs; this may have been related to the left sidedhyperesthesia, since no musculoskeletal abnormalitywas found.The nature of the episodes, the circling to the right,

and the left hyperesthesia, were consistent with a lesionof the right thalamocortex. Because of the prolongedhistory, the focal nature of the neurological deficit,and the age of the patient, neoplasia was consideredthe probable cause of the signs. Hematological stud-ies, biochemical profile, urinalysis, thoracic radio-graphs, and electrocardiogram were done.

MRI and postmortem findingsFigure 4 shows a combined T1- and T2-weightedmagnetic resonance image from this case. An area ofincreased signal intensity was seen in the right cerebralhemisphere. The abnormal area was observed in fivecontiguous slices. The location (supratentorial) andsignal characteristics of this lesion were suggestive ofa meningioma. This diagnosis was confirmed on histo-pathological examination.

Case 2History and localization of lesionThis five-year-old male Boston terrier was presentedto OVC with a four week history of seizures that wereincreasing in frequency. For four weeks prior toexamination, the animal was lethargic, depressed, andexhibited abnormal behavioral patterns (house soiling,polyphasia, and pica). Neurological examinationrevealed decreased pain response on the left side ofthe face, a tendency to circle to the right, and abnormalbehavior. These signs were consistent with a rightthalamocortical lesion. The CSF analysis was sug-gestive of chronic meningoencephalitis. A tentativediagnosis of granulomatous meningoencephalitis wasposed.

Figure 5. MRI from case 2. A. T1-weighted MRI throughthe parietal lobe. TE = 25 ms, TR = 1 sec. B.T2-weighted MRI at the same level as A. TE = 100 ms,TR = 3 sec. C. T2-weighted MRI through the frontal lobe.TE = 100 ms, TR = 3 sec. Straight arrows show the edemain the right hemisphere. Curved arrows show hyperintensemass.

Can Vet J Volume 33, September 1992588

Figure 6. MRI from case 3. A. T2-weighted imagethrough frontal lobe showing area of increased signal inten-sity consistent with inflammation (arrow). B. TI-weightedimage through the brainstem showing hypointense mass atthe level of midbrain (arrow).

MRI and postmortem findingsA T1-weighted MRI showed distortion of the rightcerebral hemisphere (Figure 5A), but the lesion was

indistinguishable from normal brain. TheT2-weighted images (Figures 5B, 5C) provided thecontrast necessary to identify the lesion. A well definedhyperintense mass was seen in the right cerebralhemisphere surrounded by a large, more diffusearea of hyperintense signal compared to normaltissue. These observations were consistent with a

space-occupying lesion with surrounding edemaand/or inflammation. The abnormal area extendedfrom the frontal lobe into the parietal lobe andthalamus. Treatment with steroids for seven days anda follow-up scan showed no change. The animal was

euthanized and histopathological examination showeda neuroesthesioma with peritumoral edema andatrophy consistent with the MRI findings.

Case 3History and localization of lesionThis three-year-old female, spayed pug was presentedto OVC with a history of recurrent seizures that did

Can Vet J Volume 33, September 1992

not respond to anticonvulsant therapy. Neurologicalexamination revealed a left-sided hemiparesis withproprioceptive deficits and left-sided cortical visualimpairment. There was anisocoria, the right pupilbeing miotic. Based on clinical signs, the lesion waslocalized to the right thalamocortex with possible com-pression of the midbrain. Analysis of CSF revealed anonsuppurative inflammatory process, and treatmentwith corticosteroids has been successful in temporarilyalleviating clinical signs.

MRI findingsA T2-weighted scan showed areas of diffuse signalhyperintensity compared to normal brain in the rightcerebral hemisphere (Figure 6A). A TI-weighted setof images revealed a mass at the level of the midbrainthat was slightly hypointense compared to normaltissue (Figure 6B). The areas of increased signal wereconsistent with an inflammatory process. Based on theCSF and MRI findings, the most likely diagnosis atthis time is granulomatous meningoencephalitis.

DiscussionIn all three clinical cases presented here, the MRI find-ings were consistent with the clinical signs and CSFanalysis. In agreement with two previously reportedcases (4), the MRI findings in cases 1 and 2 also cor-related withthe histopathological findings. In case 2,the use of MRI increased the specificity of diagnosis,which, based on clinical signs and CSF analysis, wouldhave been diagnosed as a purely inflammatory disease.The abnormal areas in these three cases all had dif-

ferent TI and T2 values as compared to that ofnormal brain tissue and could therefore be readilyvisualized by manipulation of the normal MRI pulsesequences. This is not always the case. In oneretrospective review of brain tumors (5), 120o of thetumors did not display evidence of an increase in T1or T2. Similar signal intensity to normal tissue wasobserved in a variety of tumors, but most frequentlyin lipid-containing tumors, malignant melanomas,tumors containing hemorrhage, acoustic neuromas,and meningiomas. Normal appearing signal intensitymay also be seen in any mass which is highly cellularwith relatively little intersitial space, such as lymphomaand medulloblastoma. Hyperintensity on T1-weightedimages is usually indicative of fat (such as in dermoid,lipoma, and teratoma) or hemorrhage (includingnecrotic gliomas and metastases). Studies on the relax-ation characteristics of tumors have shown that T,and T2 are not diagnostic in terms of thehistopathological classification of the tumor. Manytumors exhibit similar characteristics; however, withthe use of T1- and T2-weighted imaging, accuratelesion description (solitary lesion or multifocal, intra-or extramedullary) in combination with CSF analysisand clinical history, it is often possible to speculateon the type of mass present. This requires an under-standing of the fundamental principles of MRI anda knowledge of neuroanatomy.An additional aid to diagnosis of tumor type is the

use of an intravenous contrast agent. These agents arealso mandatory for masses that have similar intensity

589

to normal tissue, since it is difficult to see their sizeand extent. In CT of the central nervous system, con-trast enhancement is also used. The contrast agent isgiven intravenously, and, if the blood-brain barrier isincomplete, or if there is increased vascularity, theagent enters the abnormal area and highlights it on thescan. The same principle applies in MRI contrastenhancement. However, the contrast agent in CT isbased on density since X-rays are being used. In MRI,the contrast agent alters the relaxation characteristicsof mobile protons in the abnormal area; that is, itshortens T1 and T2, and highlights the lesion on aTI-weighted scan. The most commonly used contrastagent is gadolinium diethylenetriamine pentaacetic acid(Gd-DTPA). This complex is routinely used in humanstudies (6), and has greatly increased the diagnosticspecificity of intracranial masses. For example, due tothe vascular nature of meningiomas, these tumors arerapidly enhanced with Gd-DTPA (7,8). This is helpfulin distinguishing this tumor from others that mimicthe appearance and location of meningiomas but areless vascular. The Gd-DTPA complex does not crossthe intact blood-brain barrier, and it allows masses tobe classified as homo- or heterogeneous. In addition,the necrotic and cystic components are seen moreclearly. The margins of enhancement also provide agross measure of tumor extension (9,10). It must benoted however, that normal central neryous systemstructures that do not have a blood-brain barrier willalso enhance with Gd-DTPA (11). These are the choroidplexus, pituitary gland, infundibulum, dura, cavernoussinuses, cortical veins, and sinus mucosa. Clinical stud-ies using Gd-DTPA are in progress at our facility.

Magnetic resonance imaging is especially useful fordifferentiating an inflammatory process from a neo-plastic mass. The ability to differentiate tumor fromperitumoral edema is also extremely important ifsurgery is the treatment choice. A number of veteri-nary institutions are now considering the purchase ofan MRI system. This could have a dramatic impacton the diagnosis and treatment of veterinary neuro-logical diseases.

AcknowledgmentsThis work was supported in part by funds from theMRI Facility, OVC, University of Guelph. We thankBrian McCaw for his photographic services in produc-ing the clinical case figures, and Winn Halina,Biomedical Sciences, OVC for providing photographsof the gross anatomical slices from the normal dog.

cvJ

References1. Daniels DL, Haughton VM, Naidich TP. Cranial and Spinal

Magnetic Resonance Imaging. An Atlas and Guide. New York:Raven Press, 1987: 5-196.

2. Edelman RR, Hesselink JR, eds. Clinical Magnetic ResonanceImaging. Philadelphia: WB Saunders, 1990: 379-652.

3. Kraft SL, Gravin PR, Wendling LR, Reddy VK. Canine brainanatomy on magnetic resonance images. Vet Radiol 1989; 30:147-158.

4. Panciera DL, Duncan ID, Messing A, Rush JE, Turski PA.Magnetic resonance imaging in two dogs with central nervoussystem disease. J Small Anim Pract 1987; 28: 587-596.

5. Mackay IM, Bydder GM, Young IR. MR imaging of centralnervous system tumors that do not display increase in Ti orT2. J Comput Assist Tomogr 1985; 9: 1055-1061.

6. Hesselink JR, Healy ME, Press GA, Brahme FJ. Benefits ofGd-DTPA for MR imaging of intracranial abnormalities.J. Comput Assist Tomogr 1988; 12: 266-274.

7. Spagnoli MV, Goldberg HI, Grossman RI, et al. Intracranialmeningiomas: High field MR imaging. Radiology 1986; 161:369-375.

8. Bydder GM, Kingsley DPE, Brown J, Niendorf HP, Young IR.MR imaging of meningiomas including studies with and withoutgadolinium-DTPA. J Comput Assist Tomogr 1985; 9: 690-697.

9. Graif M, Bydder GM, Steiner RE, Niendorf P, Thomas DG,Young IR. Contrast-enhanced MR imaging of malignant braintumors. AJNR 1985; 6: 855-862.

10. Felix R, Schorner W, Laniado M, et al. Brain tumors: Imag-ing with gadolinium-DTPA. Radiology 1985; 156: 681-688.

11. Kilgore DP, Bregar RK, Daniels DL, Pojunas KW, WilliamsAL, Haughton VM. Cranial tissues: Normal appearance afterintravenous injection of Gd-DTPA. Radiology 1986; 160:757-761.

0 . 0

*0-

Clark Cages Clark Cage Dryerv Formica * Heavy-duty construction

Lined l_ . Fully assembled. Quiet &

* SensiblyPriced* Durable

* Attractive -- -_* Easy To _ Virtually maintenance-freeAssemble . Dries animal quickly and

safely* Space-saver design

Thrifty Cage and Run Floors* Lightweight

polypropylene _ _ Forafreeplastic panels crolcoura

* Easy-to-clean price list. Cage floors elevated contact

on 3/4" legs mmitted To* 2' x 2' panels Clark Cages Inc. Ecceltencelnartgetrunflor Canada (705) 497-8900Available in e USA Texas (214) 744-5954custosiales CaPenn. (215) 922-5564custom sizes L.JCalif. (213) 689-9324

590 Can Vet J Volume 33, September 1992