7/22/2014

Principles of PET/CT
Quality Control and Calibration

S. Cheenu Kappadath, PhD, DABR, DABSNM

Department of Imaging Physics

The University of Texas MD Anderson Cancer Center, Houston, Texas

skappadath@[Link]

Educational Objectives
 Review principles of PET and PET/CT imaging
 Overview of PET performance testing
– NEMA NU-2 2001/2007 and ACR
 PET/CT Image Quality and Artifacts
 Recent advances in PET/CT

S. Cheenu Kappadath, PhD AAPM 2014 2

Principles of PET Imaging

 Positron decay physics
 PET detectors design
 PET Lines-of-Response and Sinograms
 PET event types
– prompt, true, scatter, random
 PET data processing
– normalization, attenuation, scatter, randoms
 PET/CT
 PET calibration and SUV
 PET 2D/3D acquisitions

S. Cheenu Kappadath, PhD AAPM 2014 3

1
 7/22/2014

Positron Decay
 Nuclei with low a neutron-to-proton ratio converts a
proton to a neutron via emission of positron ( +)
p = n +  + +  ; AXZ = AYZ-1 +  + + 
 Cyclotron (generator) for production of  + emitters
13
14
N(p,)11C 16
O(p,)13N C(p,n)13N 14
N(d,n)15O
15
N(p,n)15O 16
O(p,pn)15O 18 18
O(p,n) F 20
Ne(d,)18F
 Electron capture competes with positron decay
AX = AY
Z Z-1 + X-rays
– Branching Ratios
– 18F = 0.967

S. Cheenu Kappadath, PhD AAPM 2014 4

Annihilation Photons
 Energy spectrum of  + emission is continuous
  + range depends on energy
– 18F: Emax = 0.64 MeV, Range ~1 mm
– 82Ru: Emax = 3.15 MeV, Range ~2 mm
  + annihilation results in simultaneous emission of
– Two 511 keV photons
– Emitted (nearly) 180 degrees apart g: 511 keV

-
+

g: 511 keV
S. Cheenu Kappadath, PhD AAPM 2014 5

Schematic of a PET scanner

Detector
Ring
Annihilation
photon

electron

positron
Nucleus

Annihilation
photon

S. Cheenu Kappadath, PhD AAPM 2014 6

2
 7/22/2014

PET detectors

PET Detector Block

S. Cheenu Kappadath, PhD AAPM 2014 7

PET Detector Module and Rings

PET Detector Module
PET Detector Block [Link]

S. Cheenu Kappadath, PhD AAPM 2014 8

PET Scanner – Covers Off

S. Cheenu Kappadath, PhD AAPM 2014 9

3
 7/22/2014

New PET Detector Concepts

PET APD Design Siemens PET-MR

Compact, MR compatible
Poor timing resolution

PET GM-APD or SiPM

Compact, MR compatible
Good timing resolution
[Link]

S. Cheenu Kappadath, PhD AAPM 2014 10

PET Detector Ring

S. Cheenu Kappadath, PhD AAPM 2014 11

PET Signal Processing – Singles

 Detector block/module (scintillator/PMT) converts
gamma ray energy to voltage pulse
 PHA: Event trigger and energy thresholds to select
511 keV (450-650 keV)
 Location of event in crystal block/module (Anger
logic) and event time

Cherry, Sorenson, & Phelps, Physics of Nuclear Medicine, 2003

S. Cheenu Kappadath, PhD AAPM 2014 12

4
 7/22/2014

Record the Line-of-Response

 Fundamental prerequisite to PET imaging
– Photon (Singles) detection and processing
– Coincidence assessment of singles events
– Data storage and processing

S. Cheenu Kappadath, PhD AAPM 2014 13

LOR to Sinograms

Image Courtesy: Magnus Dahlbom

S. Cheenu Kappadath, PhD AAPM 2014 14

PET Coincident Events – Prompts

Bailey, Townsend, Valk, Maisey, “Positron Emission Tomography,” Springer-Verlag, 2005

S. Cheenu Kappadath, PhD AAPM 2014 15

5
 7/22/2014

PET data corrections

Measured Prompts

Correct Random

Normalize

PET image in kBq/mL Correct Deadtime loss

Correct Geometry
Scanner Calibration

Correct Axial Sensitivity Correct Scatter

Correct Attenuation

FBP/IR reconstruction Measured “True”

S. Cheenu Kappadath, PhD AAPM 2014 16

Randoms Correction
1) Randoms Rates from Singles Rates, R = 2t x S1 x S2
– Randoms are proportional to S2
– Statistically more accurate since S>>R
2) Randoms Rates from Delayed Prompts (Dt >> 10 ns)
– Real-time subtraction
– Identical deadtime characteristics to Prompts channel
– Requires more memory and statistically less accurate

Detector Detector

10 ns 10 ns
Constant Constant
Delay Fraction
Fraction
Discriminator >> 10ns Discriminator
AND

Casey and Hoffman, 1986 Delayed Prompts are Randoms

S. Cheenu Kappadath, PhD AAPM 2014 17

Normalization (uniformity correction)

Normalization
Sinogram

Pre- Post-
Normalization Normalization

Image Courtesy: Osama Mawlawi

S. Cheenu Kappadath, PhD AAPM 2014 18

6
 7/22/2014

Deadtime Correction (Siemens mCT)

Clinical
Imaging
Conditions

S. Cheenu Kappadath, PhD AAPM 2014 19

Geometry Correction
 Ring detector introduces non-uniform sampling of
LOR away from isocenter
 Geometry correction maintains uniform pixel size in
transverse plane

S. Cheenu Kappadath, PhD AAPM 2014 20

Scatter mis-positions LOR in Sinograms

Angle

Position

S. Cheenu Kappadath, PhD AAPM 2014 21

7
 7/22/2014

Model-based Scatter Estimation

detector
B Idea: To estimate the
detector number of scattered
C coincidence along a
scatter
specific LOR
P (LOR AB in figure)
S
annilation
detector
A
Ollinger, Phys. Med. Biol. 41 (1996) 153-176
Assume an annihilation at point P,
- Compute probability the photons originate along AC
- Compute the probability that the one of the photon is detected at A
- Compute the probability of second photon scattering at location S
- Compute the fraction of events scattered toward B (Klein-Nishina formula)
- The probability that the scattered photon is detected at B

Input: PET emission image, CT transmission image, LOR AB

Output: Scatter along LOR AB
S. Cheenu Kappadath, PhD AAPM 2014 22

PET Signal Attenuation

P = P1 x P2
= e-ua x e-ub
= e-u(a+b)
= e-uD

Attenuation of PET coincident

events depends on total object Nuclear Medicine: Diagnosis and therapy,
thickness only – it is independent Harbert, Eckelman, & Neumann

of source location

S. Cheenu Kappadath, PhD AAPM 2014 23

CT-based Attenuation Correction

m x ( ECT )  m w ( ECT )
HU x  1000
mw ( ECT )
 HU x 
m x ( ECT )  1    m w ( ECT ) K
 1000 
 HU x   m w ( ECT ) m x ( E ) 
m x ( E )  1    mw ( E )    
 1000   m x ( ECT ) mw ( E ) 

Energy (keV)
 Photon energies different between
CT and SPECT
0.200
 K≈1 for Compton Scatter
Attenuation at 511keV

dominates low Z at ECT (low HU)

0.150
 K≠1 for Photoelectric pertinent
for high Z at ECT (high HU)
0.100
 HU-to-m transform is piece-wise
linear (bi-modal) 0.050
LaCroix et al., IEEE TNS 41, 1994
Kinahan et al, Med Phys 25, 1998
0.000
-1000 -500 0 500 1000
CT number @ 140kVp
S. Cheenu Kappadath, PhD AAPM 2014 24

8
 7/22/2014

PET/CT w/ and w/o AC

CT PET w/o PET with Fused PET/CT
CT-AC CT-AC

Image Courtesy: Osama Mawlawi

S. Cheenu Kappadath, PhD AAPM 2014 25

Role of CT in PET/CT
Two functions for CT
as part of PET/CT
CT Dose
Requirement
Anatomic
Higher
Localization
AC of PET (Diagnostic )

Loss of anatomic and

morphologic information Moderate

Ultra-low
Loss of PET accuracy (CT-AC only)
from incorrect CT-AC

S. Cheenu Kappadath, PhD AAPM 2014 26

PET Scanner Calibration

 Perform PET scan with low known activity
– Low scatter and deadtime conditions
– Uniform cylinder – simple attenuation correction
 Convert PET true count rate (cps) into activity
concentration (Bq/mL)
𝐵𝑞/𝑚𝐿
 PET Standard Uptake Values 𝐵𝑞/𝑚𝑔

S. Cheenu Kappadath, PhD AAPM 2014 27

9
 7/22/2014

PET Calibration Phantoms

NIST traceable F-18 STD

“S” vial geometry

S. Cheenu Kappadath, PhD AAPM 2014 28

2D versus 3D PET

Detector Center Plane

D = 0; Planes = 1; Images = 16

Detector in-between Planes

D = 1; Planes = 2; Images = 15
D = 2; Planes = 3; Images = 14
D = 3; Planes = 4; Images = 13
Cherry, Sorenson, & Phelps, Physics of Nuclear Medicine, 2003

S. Cheenu Kappadath, PhD AAPM 2014 29

PET Sensitivity NEMA (GE DRX)

S. Cheenu Kappadath, PhD AAPM 2014 30

10
 7/22/2014

2D versus 3D PET
 2D PET: Collimation septa present between detector
planes in axial direction
– Reduces scatter; Uniform AX sensitivity (1 cm bed overlap)
 3D PET: No collimation present except at end of ring
– Sensitivity 3D > 2D  lower activity needed
– Randoms & Scatter 3D > 2D; Improvements in modeling of
the random and scatter events  Standard Acq. mode
– Triangular AX sensitivity profile (~50% detector overlap)

3D PET: Higher Sensitivity +

Greater No. of Beds

2D PET: Lower Sensitivity +

Fewer No. of Beds
S. Cheenu Kappadath, PhD AAPM 2014 31

PET data corrections

Measured Prompts

Correct Random

Normalize

PET image in kBq/mL Correct Deadtime loss

Correct Geometry
Scanner Calibration

Correct Axial Sensitivity Correct Scatter

Correct Attenuation

FBP/IR reconstruction Measured “True”

S. Cheenu Kappadath, PhD AAPM 2014 32

SAM Question 1
The attenuation of PET coincident events
emitted from the patient depends on the:

38% A. Patient diameter or size

33% B. Location of annihilation event in the patient
19% C. Radiopharmaceutical administered
10% D. CT scan technique

S. Cheenu Kappadath, PhD AAPM 2014 33

11
 7/22/2014

SAM Question 1: Answer

 The attenuation of PET coincident events emitted
from the patient depends on the:
A. Patient diameter or size
B. Location of annihilation event in the patient
C. Radiopharmaceutical administered
D. CT scan technique

 Answer: A – Patient diameter or size

 Reference: SR Cherry, JA Sorenson, ME Phelps, “Physics in Nuclear Medicine, 3rd Edition,”
Saunders Elsevier, 2003

S. Cheenu Kappadath, PhD AAPM 2014 34

SAM Question 2
The well counter calibration for a PET scanner
is used to:

15% A. Correct for variations in image uniformity

23% B. Correct for variations in detector gains
15% C. Correct for differences in detector coincidence timing
8% D. Convert count rate (cps) to activity concentration
(kBq/mL)

S. Cheenu Kappadath, PhD AAPM 2014 35

SAM Question 2: Answer

 The well counter calibration for a PET scanner is used
to:
A. Correct for variations in image uniformity
B. Correct for variations in detector gains
C. Correct for differences in detector coincidence timing
D. Convert count rate (cps) to activity concentration
(kBq/mL)

 Answer: D – Convert count rate (cps) to activity

concentration (kBq/mL)
 Reference: SR Meikle, RD Badawi, “Quantitative Techniques in PET,” in Positron Emission
Tomography, eds. DL Bailey, DW Townsend, PE Valk, and MN Maisey, Springer-Verlag
(London), 2005

S. Cheenu Kappadath, PhD AAPM 2014 36

12
 7/22/2014

PET NEMA NU2-01/07 (2D/3D)

NEMA Tests
 Spatial Resolution
 Sensitivity
 Scatter Fraction/Count
Rate Performance
 Image Quality
 Accuracy of correction for
count losses and randoms
Daube-Witherspoon M. et al JNM,
43(10) 1398-1409, 2002

S. Cheenu Kappadath, PhD AAPM 2014 37

NU2-01/07 Spatial Resolution Setup

 Point Sources are located at (0,1), (0,10), (10,0) cm
 6 mCi/cc, Capillary tube sources ~ 1 mL active volume
 Reconstruct FBP, 256x256 matrix, 25 cm FOV, apply
all correction but no filtration

S. Cheenu Kappadath, PhD AAPM 2014 38

NU2-01/07 Spatial Resolution

S. Cheenu Kappadath, PhD AAPM 2014 39

13
 7/22/2014

NU2-07 Spatial Resolution

S. Cheenu Kappadath, PhD AAPM 2014 40

NU2-01/07 Sensitivity Setup

 Detected coincidence count rate per unit activity in FOV
 250 mCi in ~70 cm 2.4 mL line source  5 Aluminum Sleeves
 Sensitivity measured with increasing amounts of attenuating
material and extrapolated to no attenuation

Placed at
isocenter

70 cm

S. Cheenu Kappadath, PhD AAPM 2014 41

NU2-01/07 Sensitivity: 2D @ R=0

S. Cheenu Kappadath, PhD AAPM 2014 42

14
 7/22/2014

NU2-07 Sensitivity: 3D @ R=0

S. Cheenu Kappadath, PhD AAPM 2014 43

NU2-01/07 SF & CR Performance Setup

 SF measures the sensitivity of scanner to coincidence events
caused by scatter
 CR measures the performance of the PET scanner across a
range of radioactivity levels
 ~40 mCi (3D) in 70 cm 4.8 cc line source

• Dynamic data acq. as

4x15min and 14x25min
with 25 min delays
• Total time is ~13 hr 70 cm
• Analysis performed
on sinograms with no
corrections applied

S. Cheenu Kappadath, PhD AAPM 2014 44

NU2-07 SF & CR Performance: 3D

 Peak NEC Rate: >100 kcps @ =< 30 kBq/mL

S. Cheenu Kappadath, PhD AAPM 2014 45

15
 7/22/2014

Deadtime Correction Accuracy

S. Cheenu Kappadath, PhD AAPM 2014 46

NU2-01/07 Image Quality Setup

 IEC Phantom: ~0.2mCi/mL background; ~0.8mCi/mL sphere
 ~5 mCi in the scatter phantom
 Clinical protocol used for data acquisition and reconstruction
 Draw ROIs on spheres and background regions

S. Cheenu Kappadath, PhD AAPM 2014 47

NU2-01/07 Image Quality

S. Cheenu Kappadath, PhD AAPM 2014 48

16
 7/22/2014

PET ACR Image Quality

 Specific instructions for phantom preparations
 Clinical protocol used for data acquisition and reconstruction
 Draw ROIs on cylinders and background regions

Cylinders Background
S. Cheenu Kappadath, PhD AAPM 2014 49

PET ACR Phantom Images

1. Image contrast and quantitation – cylinder SUV
2. Uniformity and artifacts – uniform section
3. Spatial resolution – cold rods

1 2 3

25, 16, 12, 8 mm 12.7, 11.1, 9.5, 7.9 mm

S. Cheenu Kappadath, PhD AAPM 2014 50

Iterative Reconstruction (18FDG PET/CT)

Mean AC, SBR: 5-to-1, Sphere ID: 37 mm
1.00

0.95
Relative AC (Measured/True)

0.90

0.85

0.80 Subsets 1
SUV
0.75

0.70
Subsets 3
Subsets 15 mean
Subsets 21
Subsets 45
0.65
Subsets 63
0.60
0 5 10 15 20 25 30 35 40 45
Max AC, SBR: 5-to-1, Sphere ID: 37 mm
1.40 Iterations

1.35
Relative AC (Measured/True)

1.30

R. Boellaard, JNM 50, 11S-20S, 2009 1.25

1.20
SUV
1.15
max
1.10

1.05

1.00
0 5 10 15 20 25 30 35 40 45
Iterations
SBR: 10-to-1
2000

1600
Background Std-Dev

1200
Subsets 1
Image
Subsets 3
800
Subsets 15
Subsets 21
Noise
400 Subsets 45
Subsets 63
0
0 5 10 15 20 25 30 35 40 45
Iterations

Kappadath et al., IEEE-MIC , M26-220, 2007

S. Cheenu Kappadath, PhD AAPM 2014 51

17
 7/22/2014

Partial Volume Effect

 Arises from the effects of finite spatial resolution on
the reconstructed PET activity distribution
– Smears the activity distribution
– Lower signal for object size smaller than 2s

The partial-volume effect Recovery coefficient versus object size

Cherry, Sorenson, & Phelps, Physics of Nuclear Medicine, 2003

S. Cheenu Kappadath, PhD AAPM 2014 52

PET/CT: motion mis-registration

Image Courtesy: Osama Mawlawi

S. Cheenu Kappadath, PhD AAPM 2014 53

PET versus CT FOV registration

Bailey, Townsend, Valk, and Maisey, “Positron Emission Tomography,” Springer-Verlag, 2005

S. Cheenu Kappadath, PhD AAPM 2014 54

18
 7/22/2014

Truncation Artifacts

SUV max changed

from 3.25 to 6.05

Image Courtesy: Osama Mawlawi

S. Cheenu Kappadath, PhD AAPM 2014 55

Metal and CT Contrast Artifacts

Image Courtesy: Osama Mawlawi

S. Cheenu Kappadath, PhD AAPM 2014 56

SAM Question 3
All of the following affect PET image quality
except:

7% A. Reconstruction parameter
14% B. Scan duration
14% C. CT scan technique
7% D. Patient size

S. Cheenu Kappadath, PhD AAPM 2014 57

19
 7/22/2014

SAM Question 3: Answer

 All of the following affect PET image quality except:
A. Reconstruction parameters
B. Scan duration
C. CT scan technique
D. Patient size

 Answer: C – CT scan technique

 Reference: O Mawlawi, SC Kappadath, T Pan, E Rohren, HA Macapinlac, “Factors affecting
quantification in PET/CT imaging,” Current Medical Imaging Reviews 4, 34-45, 2008

S. Cheenu Kappadath, PhD AAPM 2014 58

SAM Question 4
The minimum CT dose appropriate for PET/CT
examinations are constrained by:

0% A. Accuracy of CT-based attenuation correction

14% B. Radiologist preference for CT image quality
10% C. Equalize the CT dose to the PET dose
7% D. Accuracy of PET scatter correction

S. Cheenu Kappadath, PhD AAPM 2014 59

SAM Question 4: Answer

 The minimum CT dose appropriate for PET/CT
examinations are constrained by:
A. Accuracy of CT-based attenuation correction
B. Radiologist preference for CT image quality
C. Equalize the CT dose to the PET dose
D. Accuracy of PET scatter correction

 Answer: B – Radiologist preference for CT image

quality
 Reference: FH Fahey, MR Palmer, KJ Strauss, RE Zimmerman, RD Badawi, ST Treves,
“Dosimetry and adequacy of CT-based attenuation correction for pediatric PET: Phantom
study,” Radiology 243, 96–104, 2007

S. Cheenu Kappadath, PhD AAPM 2014 60

20
 7/22/2014

Recent Advances in PET/CT

 Recent advances
– TOF PET
– PSF modeling
– Extended axial FOV
– Gating for motion correction

 More recent advances

– Continuous bed motion (Siemens FlowMotion)
– Digital detectors (Philips Vereos)
– Regularized reconstruction (GE [Link])

S. Cheenu Kappadath, PhD AAPM 2014 61

Time-of-Flight PET

Probability along LOR

∆𝑡
Dt (ps) Dx (cm) ∆𝑥 = 𝑐
2
600 9
100 1.5 𝐷𝑜𝑏𝑗
𝑆𝑁𝑅𝑇𝑂𝐹 ≅ 𝑆𝑁𝑅𝑛𝑜𝑛−𝑇𝑂𝐹
0.33 0.5 ∆𝑥

S. Cheenu Kappadath, PhD AAPM 2014 62

TOF PET Image Quality

TOF PET Non-TOF PET

Image Courtesy: Osama Mawlawi

S. Cheenu Kappadath, PhD AAPM 2014 63

21
 7/22/2014

PSF Resolution Modeling

Lee et al., PMB 49, 2004

w/o w/

Pecking et al., Clin. Exp. Metastasis 29, 2012

 Goal is to improve image

w/o w/
quality, contrast, and
quantitative accuracy
 SharpIR (GE)
 TrueX (Siemens)
 Phillips 
S. Cheenu Kappadath, PhD AAPM 2014 64

Improvements in PET Image Quality

S. Cheenu Kappadath, PhD AAPM 2014 65

Advantages of Extended Axial FOV

Sensitivity

FOV
 Fewer bed positions for same axial coverage
 Increased sensitivity  time/bed or counts/time
 Net reduction in imaging time (or administered
activity) for comparable image quality
Image courtesy: D Townsend
S. Cheenu Kappadath, PhD AAPM 2014 66

22
 7/22/2014

SAM Question 5
The main advantage of a TOF PET scanner
over a non-TOF PET scanner is:

10% A. Higher intrinsic spatial resolution

10% B. Higher image contrast-to-noise ratio (CNR)
7% C. Higher count-rate performance
7% D. Lower number of detector elements needed

S. Cheenu Kappadath, PhD AAPM 2014 67

SAM Question 5: Answer

 The main advantage of a TOF PET scanner over a
non-TOF PET scanner is:
A. Higher intrinsic spatial resolution
B. Higher image contrast-to-noise ratio (CNR)
C. Higher count rate performance
D. Lower number of detector elements needed

 Answer: B – Higher image contrast-to-noise ratio

(CNR)
 Reference: M Conti, “Focus on time-of-flight PET: the benefits of improved time resolution,”
EJNMMI 38, 1147-1157, 2011

S. Cheenu Kappadath, PhD AAPM 2014 68

Gating and List Mode

 Motion smears PET signal and reduced intensity
– PET is motion averaged therefore use (motion) average CT
 Trigger to sort PET data into bins to correct for organ
motion – cardiac or respiratory gating

SUV = 5.0 SUV = 8.5

Image courtesy: Tinsu Pan
S. Cheenu Kappadath, PhD AAPM 2014 69

23
 7/22/2014

Gated 4D PET and 4D CT Acquisition

Trigger Trigger
1 8 1 8
2
2 7
7
3 3 6
6
4 4 5
5

time

Bin 1 Bin 8
• Prospective fixed forward time binning
• Single FOV Gated PET and Gated CT
• User defined number of bins and bin duration
• Images will be noisy unless acquired for longer durations
Image Courtesy: Tinsu Pan

Motion Correction Software

 Goal is to improve image quality, contrast, and
quantitative accuracy – respiratory motion
 [Link] (GE): Phase-matched 4D PET/CT
 [Link] (GE) and [Link] (Siemens): Use PET data
from end-expiration when motion is low
 Other vendors also have 4D PET solutions

image courtesy: Siemens

S. Cheenu Kappadath, PhD AAPM 2014 71

Continuous Bed Motion

One-Size-Fits-All Siemens FlowMotion

2’
Hi-Rez

0.8 mm/s
2’
HD•Chest

2’
0.5 mm/s
2’

2’ 0.8 mm/s
Speed
Higher

2’ 2.0 mm/s

image courtesy: Siemens

S. Cheenu Kappadath, PhD AAPM 2014 72

24
 7/22/2014

Continuous Bed Motion

 Siemens FlowMotion mCT scanner

image courtesy: Siemens

S. Cheenu Kappadath, PhD AAPM 2014 73

Regularized Reconstruction – GE [Link]

PSF TOF+PSF QC+PSF QC+TOF+PSF

77 years male with follicular lymphoma, 80 kg, 25 BMI, 9.4 mCi, 60 min post injection

S. Cheenu Kappadath, PhD AAPM 2014 74

Fully Digital PET/CT – Philips Vereos

 LYSO crystals + SiPM  Fully digital detectors
– Fast and high sensitivity
 TOF, PSF modeling, 4D capability

image courtesy: Philips

S. Cheenu Kappadath, PhD AAPM 2014 75

25
 7/22/2014

References
 SR Cherry, JA Sorenson, ME Phelps, “Physics in Nuclear
Medicine, 3rd Edition,” Saunders Elsevier, 2003
 DL Bailey, DW Townsend, PE Valk, and MN Maisey, “Positron
Emission Tomography,” Springer-Verlag (London), 2005
 M Conti, “Focus on time-of-flight PET: the benefits of
improved time resolution,” EJNMMI 38, 1147-1157, 2011
 O Mawlawi, SC Kappadath, T Pan, E Rohren, HA Macapinlac,
“Factors affecting quantification in PET/CT imaging,” Current
Medical Imaging Reviews 4, 34-45, 2008
 FH Fahey, MR Palmer, KJ Strauss, RE Zimmerman, RD Badawi,
ST Treves, “Dosimetry and adequacy of CT-based attenuation
correction for pediatric PET: Phantom study,” Radiology 243,
96–104, 2007

S. Cheenu Kappadath, PhD AAPM 2014 76

26