Medical ImagingSource: uhrad.com.
Medical ImagingSeeing Through the Human Bodyg g y
El A li iElsa AngeliniDepartment TSIENST
1
ENST
It all started in 1845 with the first X-ray image
2Roentgen’s wife Von Kolliker
History of Medical ImagingX
Ray • 1845: Discovery of X-ray by Wilhelm Conrad Röntgen, german physicist.
• 1972: First scanner set up by Allan Mc Cornack et Godfrey N. Hounsfield, Nobel prize in 1979 for this work.
US • 1915: Ultrasound propagation principles for SONAR (SOund NAvigation Ranging).
• 1955: First echography by Inge Edler (1911-2001), swedish cardiologist.
MR
I • 1945: Discovery of nuclear resonance under a magnetic field by Edward Purcell and Felix Bloch, both nobel prize winners in 1952.
• 1973: First MRI image on an animal by the american chemist Paul Lauterbur. P. Lauterbur and P. Mansfield won the Nobel Prize for Physiology or Medicine in 2003 for the development of MRI.
lear
the development of MRI. • 1980: Spectroscopy by magnetic resonance.
• 1934: Discovery of natural radioactivity by Henri Becquerel, Pierre et Marie Curie. Discovery of artificial radioactivity by Irène et Frédéric Joliot-Curie
3
Nuc
l y y y• 1990: Nuclear medicine development with scintigraphy and positron emission
tomography.
Terminology• Transmission: Photons passing through the body• Absorption: Partial or total absorption of energy in the patient• Scatter: Radiation diverted in a new direction
• Transaxial - plane normal to a vector from head to toe. Source: Utrecht University, NL
p• Coronal - plane normal to a vector from front to back • Sagittal - plane normal to a vector from left to right. • Oblique – otherwise
4
• Oblique – otherwise.
Imaging Data Types• 2D:
– A single slice.A single slice. • 3D:
A series of parallel slices with uniform– A series of parallel slices with uniform spacing.
– A series of slices with varying spacing– A series of slices with varying spacing and/or orientation.
– A volumetric data set
5
A volumetric data set.
Imaging Support
Digital DisplayFilms
6
(soft copy)(hard copy)
Imaging Modalities
• X-Ray:– TraditionalTraditional – Computerized Tomography
• Ultrasound• Ultrasound• Magnetic Resonance Imaging (MRI)• Nuclear Imaging
7
Conventional X-Ray
8
Conventional X-RayX ra sPhysics Principles
• Accelerating & decelerating electrons generate electromagnetic radiation
• Bombardment of a target material with a
X-rays E ≈ 5-150keV
Physics Principles
• Bombardment of a target material with a beam of fast electrons
• Electrons are emitted thermally from a heated cathode (C) and are accelerated t d th d t t (A) b th li d
CAe-
toward the anode target (A) by the applied voltage potential V (~kV).
• Electromagnetic radiation is produced by the quick deceleration of the electrons
+-V = 5-150kV
qwhen hitting the target (bremsstrahlung). – ~99% of energy converted into heat.– Few generated X-ray photons with
wavelength set by the amount of energy E1
hν
9
wavelength set by the amount of energy loss E=hν=E2-E1
E1
E2Nucleus
Conventional X-RayI F ti
• X-ray beam goes though th ti t ith d
Image Formation
the patient with decrease energy, function of the absorption pattern inside Primary X-raysabsorption pattern inside the body.
patient
Primary X rays
collimator
10
Screen + detector
Conventional X-RayI C t
• Image captured on h h ith
Image Capture
phosphor screen, with conversion to visible light.
Primary X-• Recording on film
(negative image) patient
lead
Primary Xrays
• Display on video monitor (positive image)
lead
11
(p g )detector
Conventional X-RayS t D iSystem Design
cathodescatters
lead
VERRE BOMBARDE
CONE PLOMBE
HUILE oil
Optical field
12Optical field
Conventional X-RayS t D iSystem Design
13
Conventional X-Ray• Advantages:
• Fast.• Inexpensive.
• Disadvantages:• Ionizing radiation involved.• Projection images.• Poor blood vessel and soft-tissue
contrasts.
14
Conventional X-Ray
• Applications:– Lungs, breasts, GI track.Lungs, breasts, GI track.– Blood vessels (contrast agent).– Bones jointsBones, joints.
15
Source: http://www.szote.u-szeged.hu/Radiology/Anatomy/
Conventional X-RayExamples: Bones
1 Frontal sinus 11. Mastoid cells12 N h
1. Bifid spinous process of C32. Superimposed articular processes3 Uncinate processes
1. Frontal sinus2. Ethmoidal sinus3. Sphenoidal sinus4. Maxillary sinus5. Anterior clinoid processes6. Hypophyseal fossa7 Posterior clinoid processes
12. Nasopharynx13. Angle of mandible14. Anterior arch of the atlas15. Dens of axis16. Posterior arch of the atlas17. Internal occipital protuberance
16
3. Uncinate processes4. Air filled trachea5. Transverse process of C76. Transverse process of T17. 1st rib8. Clavicle
7. Posterior clinoid processes8. Clivus9. Great density of the petrouspart of the temporal bone10. External acoustic meatus
A. Coronal sutureB. Lambdoid sutureC. The grooves for the branches of the middle meningeal vessels
Examples: LungsConventional X-Ray
Patient with hemo-pneumothorax Patient with pacemaker
Pleurall line
air + fluid
17Source: www.mypacs.net
air + fluid
Examples: MammographyConventional X-Ray
p g p y• Low kV can be used (little
attenuation) => sufficient photoelectric absorption tophotoelectric absorption to differentiate between normal tissue and pathology
• Dose must be limited• Dose must be limited.• Breast must be compressed
– To avoid motion.– To minimize focus/film distance.– To have equal tissue thickness.
18
Examples: FluoroscopyConventional X-Ray
Examples: Fluoroscopy• Real-time X-ray imaging.
Direct is ali ation on screen• Direct visualization on screen.• Used for guidance of interventions
(mainly vascular and orthopedic).
Ex: Fluoroscopic guidance withEx: Fluoroscopic guidance with umbilical catheter injecting contrast agent in the small bowel
19
bowel.
Source: www.mypacs.net
Examples: FluoroscopyConventional X-Ray
Examples: Fluoroscopy
patient
C-arm
Source: Utrecht University NL
20
Source: Utrecht University, NL
cathode
C t i d T hComputerized Tomography (CT)(CT)
21
CTPh i P i i lPhysics Principles
• Same as X-ray.
22
CTI F ti
• Same as X-ray.Image Formation
23
CTI C tImage Capture
• Generation of a sliced view of body interior (“T”)
Measure = integral attenuation
of body interior ( T ) (cf. tomos = slice in Greek).• Computational intensive
µ(x,y) ?
• Computational intensive image reconstruction (“C”).
Patient’s body
X
24
X-ray source
CTCTI C tImage Capture
• Translation and rotation of integral measure of absorption.
• Backprojection
∫⎛ ⎞
I = Ioexp − µ(x, y)dlL∫
⎛
⎝ ⎜
⎞
⎠ ⎟
=> µtotal = µ(x,y)dl∫ = −ln I⎛ ⎜
⎞ ⎟
25
> µtotal µ(x,y)dl
L∫ ln
I0⎝ ⎜
⎠ ⎟
∫⎛ ⎞
g(s,θ)CT
Image Capture: BackprojectionI = Ioexp − µ(x, y)dl
L∫
⎛
⎝ ⎜
⎞
⎠ ⎟
I⎛ ⎜
⎞ ⎟
x’y
sy’
g ≡ ln Io
I⎛ ⎝ ⎜
⎞ ⎠ ⎟ (Signal)
θs
µ(x,y)xunknown
absorption g(s,θ ) = µ(x, y)dlL∫
The Radon transform g(s,θ) of a function µ(x,y) is defined as its line integral along a line inclined at an angle θ from
26
as its line integral along a line inclined at an angle θ from the y-axis and at a distance s from the origin.
CTImage Capture: Backprojection
g(s,θ ) = µ(x, y)dlL∫ g(s,θ)L
• The Radon transform maps the spatial domain (x,y) to the domain (s,θ).
x’y
sy’
domain (s,θ).• Each point in the (s,θ) space corresponds to a line in the spatial domain (x,y). x
θµ(x,y)
unknownunknownabsorptiong(s,θ ) = µ(x, y)δ (x cosθ + y sinθ − s)dxdy∫∫
= µ(s cosθ − y' sinθ ,s sinθ + y' cosθ)dy'∫
27
µ( y , y ) y∫
2D-frequency domain of µ(x,y)
CTImage Capture: Backprojection
2D-space domain of µ(x y)
µ( ,y)Radon Transform & Projection Theorem (Center slice theorem)
2D-space domain of µ(x,y)
g(s,θ) sy
θ
ωy
θ
s1D-Fourier transform
F( ( ))x
θ
µ(x y)ωx
θF(g(s))
28
µ(x,y)
CTImage Capture: Hounsfield Units
• Relates the linear attenuation coefficient of a local region µ to the linear attenuation coefficient of water, µW @ 70 keV (=Eeff).Based on meas rements ith the original EMI scanner in ented b• Based on measurements with the original EMI scanner invented by Hounsfield.
HU = 1000 •µ − µw
µ w
29• Assign calibrated values to gray scale of CT images.
CTS t D i• 1st generation: Original resolution:
80 × 80 pixels (ea. 3 × 3 mm2),
System DesignTranslation & Rotationp ( ),
13mm slice thickness, 5min scan time, 20min reconstruction.
• 2nd generation: reduced number of view angles ⇒ scan time ~30 s.
Translation & Rotation
• 3rd generation: scan time ~seconds (reduced dose, motion artifacts). Reconstruction time only )~seconds.
yRotation
CTS t D i
• 4th generation: scan time, reconstruction time ~sec
System Design
reconstruction time sec.
only Rotation ofonly Rotation of sources
• Spiral: Continuous linear motion of patient table during multiple scans. I d lIncreased coverage volume per rotation.
CTExample: Image quality
20001975
32
128x128 pixels, 1-4 hours acquisition, 1-5 days computation
512x512pixels, 0.35 sec aquisition, <1sec computation
CTExample: Lungs
33
CTExample: Lungs
Ex: CT of right lung with active pulmonary tuberculosis: shows centrilobular lesions (nodules
b hi li t t 2 4 i di t )or branching linear structures 2-4 mm in diameter) in and around the small airways.
Source: www.mypacs.net
34
CTExamples: Abdomen
35Diagnosis of appendicitis
CTExample: Abdomen
TRA CTScout CT Coronal MPR
Virtual d
Source: www.mypacs.net
endoscopy
36
CTExample: Blood vessels
37
CTExample: Blood Vessels
38
CT
39Vitrea©
• Calcium ScoringCT
T t l S S
Calcium Scoring
Hn x-factor(Agatston Scoring)
Total Score = S
(Agatston Scoring)
130-199 1
200 299 2200-299 2
300-399 3
>400 4
40
>400 4
Area = 15 mm2
Peak CT = 450Score = 15 x 4 = 60
Area = 8 mm2
Peak CT = 290Score = 8 x 2 = 16
Ultrasound
41
UltrasoundPh i P i i lPhysics Principle
• Traveling pressure waves through a medium.
42
UltrasoundPh i P i i l
20 Hz 20 kHz 1 MHz 20 MHz
Physics Principle
20 Hz 20 kHz 1 MHz 20 MHzinfra-sound
audible sound ULTRASOUND
DiagnosticDiagnosticUltrasound
Frequency f (Hz)101 102 103 104 105 106 107
flight ~ 1015
43
q y ( )flight 10fxray ~ 1018
UltrasoundI F ti
• Sound waves propagate th h th b d
Image FormationN
e
through the body. • At each interface, part is
fl t d t i
ear Field
reflected, part is transmitted.R fl t d
dFar F• Reflected waves are
recorded back.
Field
44Source: The essential physics of Medical Imaging
Ultrasound PulsetransducerUltrasound
I F ti
Echo
Image Formation
Echo
DD
EchoTISSUE D = c×Time/2
c = 1 54 mm/µs
45
c = 1.54 mm/µs
UltrasoundImage Captureg p
• Recorded signals:– Specular reflection, Diffuse
scattering, Rayleigh scattering• Frame rate (FR) =number of
times per second a sweep of the ultrasound beam is performed.– The higher the frame rate, the
T_line = 2×D/c
T_frame = N ×Tline = 2 N D/
gbetter the temporal resolution.
– Limited by:• Time delay from maximum depth.
2×N×D/c
FR_max = 1/ Tframe
46
• Speed of sound.
Piezoelectric ElementA ti I l t
UltrasoundS t D i Acoustic InsulatorSystem Design Transducer
Backing Material
ElectricConnector
Matching Layer
Backing MaterialPhased array transducer
Di lDisplay
47
Ultrasound• Advantages:
– Portable.– Inexpensive.– Good spatial resolution (~1mm).– High temporal resolution.
• Disadvantages:– Poor image quality.– High level of noise.
48
UltrasoundS t D iSystem Design
A-modeamplitude displayamplitude display
B-modeB i ht di l
49
Brightness display(Gray-scale)
UltrasoundS t D iSystem Design
A-mode B-mode
50
UltrasoundApplications
• ObstetricH t
pp
• Heart• GI track (liver, gall bladder, pancreas)• Urinary track (bladder, kidneys) • Genital organs (prostate, testicles,Genital organs (prostate, testicles,
ovaries, uterus). • Blood vessels (Doppler)
51
• Blood vessels (Doppler).
Example: Blood vesselsUltrasound
– Establish number of fetus.Check position of placenta
Example: Blood vessels
– Check position of placenta. – Determine fetal age and development.– Check for congenital defects.Check for congenital defects.– Check fetus position.
52
Example: ObstetricUltrasound
1961: First image of fetus 1990: First 3D visualization of fetus
Example: Obstetric
fetus
53
Example: HeartUltrasound
- Valve diseases.- Cardiac function (pumping efficiency).
Bl d l t i di h b
Example: Heart
- Blood clots in cardiac chamber.
54
UltrasoundExample: HeartExample: Heart
55• Real-Time 3D Ultrasound
• Biopsy guidance. Examples: AbdomenUltrasound
p y g• Diagnosis of abdominal pains, kidney stones, inflamed appendicitis. kidney
Examples: Abdomen
appendicitis
56
liver
Example: Blood vesselsUltrasound
• Artery blockage from blood clots.• Arteriosclerosis (plaques).• Stent position
Example: Blood vessels
• Stent position
57
UltrasoundUltrasoundExample: Blood vessels
• DopplerExample: Blood vessels
58
Magnetic Resonance ImagingMagnetic Resonance Imaging(MRI)
59
MRIPh i P i i l
• Nuclei with magnetic dipoles (H)Physics Principle
No magnetic field Magnetic field
(H). • Magnetic field B0 makes nuclei precess along its
directiondirection.• The precession rateis the Larmor frequency.
• f = γ B (43×1 5 = 64 MHz for• fL = γ B0 (43×1.5 = 64 MHz for Hydrogen into a 1.5T magnet).
B0
z
60
0
Source: Hielscher, Columbia Univ.
MRIImage Formation • Nuclei with magnetic dipole moment µ,g
En
erg
y
mz = +1/2
B0
z
g p µ,and angular momentum S.• Net magnetization M = density of µ.
• Tip M away from B ⇨ Precession atE e y
mz = -1/2
Spin flip
∆Ε = γΒ0
• Tip M away from B0 ⇨ Precession at frequency γB0, producing a measurable RF magnetic field.
Z
J or µ or MB0
• For every 1,000,000 nuclei in upper state|B
transition
Y
For every 1,000,000 nuclei in upper state (mz=+1/2) there are 1,000,001 nuclei in lower state (mz=-1/2).
• This difference is enough to result in i t ti ti f t i l
B0|
61
Y
X
|B0|×γ×tmacroscopic net magnetization of material.
Image CaptureMRI
• As the magnetization precesses it creates its own RF magnetic field.
• This field is much smaller than the exciting RF field
Z
J or µ or MB0
exciting RF field.• It can be detected with a standard radio
receiver (e.g. RF coils used to tip M away from B0)
J or µ or M
from B0)• The resulting signal from precession is
called the Free Induction Decay or FID.Y
X
|B0|×γ×t
X
62
MRISystem Design• Main Magnet
– High, constant,Uniform Field, B
B0B0.
• Gradient CoilsProduce pulsed linear gradients in
B0– Produce pulsed, linear gradients in this field: Gx, Gy, & Gz
• RF coils B1RF coils– Transmitter: B1 Excites NMR signal
to generate Free Induction Decay ( FID)
B0
63
FID).– Receiver: Senses FID.
MRISystem Design
Magnet + shielding + cooling systems
gantry
64
g y
MRI• Advantages:
– High image quality.– Good spatial resolution (~1mm).– All tissues (except lung).
• Disadvantages:– Very expensive scanners.– Some patients cannot be imaged
(pacemaker, claustrophobia).
65
MRIApplications
• Brain (Multiple sclerosis, Alzheimer’s disease epilepsy tumors)
pp
disease, epilepsy, tumors) • Spine• Muscles, tendons. • Blood vessels: angiography (strokes,
aneurism)• Heart• Heart• soft tissues
66
Example: HeartMRI
• CINE MRIExample: Heart
67
Courtesy of Robert R. Edelman
Example: HeartMRI
Example: Heart
68MRI for myocardial perfusion
Example: Blood vesselsMRI
M ti R
Example: Blood vessels
Magnetic Resonance Angiography
69
Example: Blood vesselsMRI
Example: Blood vessels
70Aorta
Example: BrainMRI
71
Example: BrainMRI
• Diagnosis of Alzheimer’s and Parkinson’s disease.Example: Brain
72
Example: Blood vesselsMRI
Example: Blood vessels
MR Angiography
73
Nuclear Imaging (PET & SPECT)(PET & SPECT)
74
Nuclear ImagingPh i P i i l PETPhysics Principle: PET
Nuclear imaging is the use of radioactive materials forimaging structures and function inside the bodyimaging structures and function inside the body.
75
Nuclear ImagingPh i P i i l PETPhysics Principle: PET• In unstable nucleus, (more protons than neutrons) proton converts to neutron under emission of positronconverts to neutron under emission of positron.
• Positron travels limited distance in tissue.
• Positron can combine with electron of nearby atom and emits two 511 keV X-ray photons (γ rays) that travel in opposite direction.pp
76
Nuclear ImagingI C t PETImage Capture: PET
Anger Camera
77
Nuclear ImagingI C t PETImage Capture: PET
78
Nuclear ImagingI C t SPECTImage Capture: SPECT
79
Nuclear ImagingI C t SPECTImage Capture: SPECT
80
Nuclear ImagingS t D i PETSystem Design: PET
PM: photo multiplier
81
Nuclear ImagingS t D i SPECT
unknownactivity & absorption
• Coincidence occurs with simultaneous detection of
System Design: SPECTα(x,y) ,µ(x,y)
activity & absorptioncross-section
)simultaneous detection of 2 opposite detectors.
• Number of counts at the detectors gives the line
I(σ,
θ)
L1Ldetectors gives the line
integral• Use standard
t h
max max
g( , ) ( , )exp ( ,L L
s x y x ydl dlθ α µ⎛ ⎞
= −⎜ ⎟⎝ ⎠
∫ ∫
Lmax
tomography reconstruction algorithm or statistical approaches (EM)
min 1L L⎝ ⎠∫ ∫
problem is ill-posed! ⇒ different combinations of
d
82
(EM). α and µcan yield identical
measurements
Nuclear Imaging
• Advantages– Functional imaging.g g
• Disadvantages– Poor spatial resolution (~8mm)Poor spatial resolution ( 8mm).– High level of noise.
Expensive hardware– Expensive hardware.– Need of cyclotron.
83
Nuclear ImagingApplications
• HeartL
pp
• Lungs• Kidneys
Bladder• Bladder• GI Track• Endocrine system (hormones)• Endocrine system (hormones)• Immune system (white blood cells).
84
Whole BodyNuclear Imaging
Clinical History: A 71year-old male with history of lung carcinoma, now with bilateral pulmonary nodules and mediastinal
Figure 1. Whole-body gallium-67 scan showing lymphoma in the chest, the spleen, the abdomen, and the groin. The chest and spleen sites were known before the scan. The abdominal and groin uptakes were detected
pulmonary nodules and mediastinal adenopathy.Findings: Bone Scan: Increased radiopharmaceutical uptake along cortical margins of the bilateral femurs and tibias. No evidence of bony metastases
85
abdominal and groin uptakes were detected only by the scan.Source: http://www.nucmednet.com/galymph.htm
evidence of bony metastases.
Diagnosis: Hypertrophic osteoarthropathy
http://www.uhrad.com/spectarc.htm
Example: LungNuclear Imaging
SPECT scan for lung perfusion (pulmonary embolism)
86Source: Brigham & women’s Hospital
Example: HeartNuclear Imaging
• Perfusion ImagingStress exams
Example: Heart
Stress exams
Abnormal blood flow
Stress
Normal blood flow Rest
87
Example: BrainNuclear Imaging
Example: Brain
88Source: www.med.harvard.edu/AANLIB
Example: BrainNuclear Imaging
Example: Brain
89Source: www.med.harvard.edu/AANLIB
Example: BrainNuclear Imaging
PET with FDG tracer for diagnosis of brain tumorExample: Brain
90Source: V. Cunningham, GlaxoSmithKline.
Example: BrainNuclear Imaging
Example: BrainPET for in-vivo testing of new drugs
91Source: V. Cunningham, GlaxoSmithKline.
Research in MedicalResearch in Medical Imaging
92
Research in Medical Imaging• Topics• Topics
– Improve image quality.– Extraction of organ contours.– Quantitative measures on organs (volume, cardiac
function, deformations, contractility,…).– Modelization of organs.– Validation of image information for diagnosis.– Tumor detection.– Functional analysis neuroscienceFunctional analysis, neuroscience.
• Tools:– Computers, mathematics, programming, biology,
physics clinical studies
93
physics, clinical studies.
Example: Denoising of 3DRX i t d
Research in Medical Imagingspine study
Original Gaussian filtering
94Adaptive Wavelet
ThresholdingWavelet
ThresholdingDenoising Application
Example: Registration of MRI d PET bd i
Research in Medical Imagingand PET abdomen images
95
Example: Segmentation of ti l b i t t
Research in Medical Imagingcortical brain structures
96
Example: Segmentation of b ti l t t
Research in Medical Imagingsub-cortical structures
Lateral ventricles
Third ventricleThird ventricle
Caudate nucleus
Thalam sThalamus
97
Example: Segmentation of b ti l t t
Research in Medical Imagingsub-cortical structures
98
Example: Diffusion Tensor MRI f B i Whit M tt Fib
Research in Medical Imagingfor Brain White Matter Fibers
99
Example: Functional Brain MRIResearch in Medical Imaging
Example: Functional Brain MRI
Motor control of fingers
100
Example: Functional MRIResearch in Medical Imaging
Example: Functional MRI
101Motor control of right hand
Example: Fetal ImagingResearch in Medical Imaging
Example: Fetal Imaging
102
Example: Fetal ImagingResearch in Medical Imaging
Example: Fetal Imaging
103
Books•Foundation of Medical Imaging, Z.H. Cho, J.P. Jones, M. Singh, John Wiley &Sons, Inc., New York 1993, ISBN 0-471-54573-254573-2.
•Principles of Medical Imaging, K.K. Shung, M.B. Smith, B. Tsui, Academic Press, San Diego 1992, ISBN 0-12-640970-6.Tsui, Academic Press, San Diego 1992, ISBN 0 12 640970 6.
•The Essential Physics of Medical Imaging (2nd Edition) ,Jerrold T. Bushberg, J. Anthony Seibert, Edwin M. Leidholdt Jr., John M. Boone, Lippincott Williams & Wilkins; 2nd edition, ISBN 0683301187.
I R i i R di l J A P k CRS P•Image Reconstruction in Radiology, J.A. Parker, CRS Press, Boca Raton, FL 1990, ISBN 0-8493-0150-5/90.
•Principles of Magnetic Resonance Imaging Zhi Pei Liang
104
•Principles of Magnetic Resonance Imaging, Zhi-Pei Liang, P.C. Lauterbur IEEE Press, New York, NY 2000, ISBN 0-7803-4723-4.
Top Related