1. Nuclear Medicine Physics
Positron Emission Tomography (PET)
Jerry Allison, Ph.D.
Department of Radiology
Medical College of Georgia

2. Medical College of Georgia And Sameer Tipnis, Ph.D.
A note of thanks to
Z. J. Cao, Ph.D.
Medical College of Georgia
And
Sameer Tipnis, Ph.D.
G. Donald Frey, Ph.D.
Medical University of South Carolina
for
Sharing nuclear medicine presentation content

3. SPECT vs PET PET SPECT (Simultaneous acquisition)
(Step-and-shoot acquisition)
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

4. SPECT vs PET imaging Attribute SPECT PET Detection Single s
Coincident s
Radionuclides
99mTc, 67Ga, 111In
18F, 82Rb, 13N, E
70 – 300 keV
511 keV
Spatial res.
 10 – 12 mm
 mm
Atten.Correction
No / Yes*
Yes
* Possible with SPECT/CT or transmission source systems
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

5. Three steps in PET imaging
1. Emission of positron by radionuclide
EC ~3.3%
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

6. Three steps in PET imaging
2. Annihilation of positron & emission of photon pair
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

7. Three steps in PET imaging
3. Coincidence detection of photon pair
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

8. Lines of response (LOR)
PET image formation t1 
 t = t1 – t2
 t < 6 (to 12) ns ?
Yes
Register as a “coincident” event  t2
Why 5 & 12ns?
Lines of response (LOR)
Positional information is gained
LOR is assigned by electronic coincidence circuitry
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

9. LORs combined to form image
Reconstructions - typically OSEM (iterative)
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

10. + emitters used in PET p  n + e+ + 
Proton-rich nuclei: positron emission
p  n + e+ + 
18F9  18O8 + e+ +  T1/2 = 110 min
15O8  15N7 + e+ +  T1/2 = 2 min
13N7  13C6 + e+ +  T1/2 = 10 min
11C6  11B5 + e+ +  T1/2 = 20 min
82Rb37  82Kr36 + e+ +  T1/2 = 73 sec

11. Two Scientists Awarded Nobel Prize In Physics For Neutrinos Discoveries OCTOBER 06, 2015
NPR: Arthur McDonald of Canada and Takaaki Kajita of Japan were awarded Nobel Prize in Physics Tuesday for discovering that subatomic particles called neutrinos can switch from one kind to another. NPR has more about the win and how it could change physics in a big way. …
Today the Nobel Prize in physics was awarded to two researchers. Takaaki Kajita, of Japan, and Arthur McDonald, of Canada, won for showing that particles called neutrinos have mass.
For neutrinos to change type, the laws of physics say they have to have mass. But the best theories at the time predicted neutrinos were weightless. In other words, these experiments showed the best theories were wrong.

12. Cosmic Gall by John Updike Neutrinos
Neutrinos, they are very small.
They have no charge and have no mass
And do not interact at all.
The earth is just a silly ball
To them, through which they simply pass,
Like dustmaids down a drafty hall
Or photons through a sheet of glass.
They snub the most exquisite gas,
Ignore the most substantial wall,
Cold-shoulder steel and sounding brass,
Insult the stallion in his stall,
And, scorning barriers of class,
Infiltrate you and me! Like tall
And painless guillotines, they fall
Down through our heads into the grass.
At night, they enter at Nepal
And pierce the lover and his lass
From underneath the bed—you call
It wonderful; I call it crass.

13. Annihilation  Energy conservation:
e- + e+ = 2  
Energy conservation:
me- = me+ = 511 keV  Eg + Eg = 2 x 511 keV
Momentum ( = m ) conservation:
pe- = pe+ = 0  pg + pg = 0 but pg  0
 2 g always in opposite directions
The two coincident g’s are detected in PET.

14. Uncertainties in annihilation
511 keV
angle divergence
180o±0.3o (18F)
positron scatters in tissue to lose energy +
511 keV
18F maximum positron range: 2.3 mm  mm FWHM resolution limitation

15. Annihilation location  Ejection location
The distance depends on the e+ initial kinetic energy and medium.
Isotope Max E Max d FWHM F MeV 2.3 mm mm C MeV 3.9 mm mm
O MeV 6.6 mm mm
Rb MeV 16.5 mm 2.6 mm
Shorter distance in a medium with higher density or higher Z

16. Aside: + endpoint energy & spatial resolution
82Rb
Emax. = 3.35 Mev
Rangerms  2.6 mm
Emax. = 0.64 Mev
Rangerms  0.2 mm
18F
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

17. Residual momentum of e+ and e-
Neither positron nor electron are at complete rest when annihilation occurs. The residual momentum causes a small angular deviation from 180.
h  × ring diameter
For D = 80 cm,
h ~ 2 mm g
LOR

18. Ultimate spatial resolution in PET
The uncertainties in annihilation (location & residual particle momentum) determine the ultimate (limiting) spatial resolution (~ 2 mm)

19. Coincidence Detection
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

20. Types of coincidences
(correct LOR assigned)
(incorrect LOR assigned)
True
Scatter
Random
True coincidences form a “true” distribution of radioactivity
Scatter & random coincidences distort the distribution of radioactivity, add to image noise, degrade image quality
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

21. Count rates P = T + R + S T = P - S - R R = t R1 R2
P = “Prompts” (count rate measured by detector pair)
T = Trues
S = Scatter
R = Randoms
R1, R2 = singles count rates in detectors 1 and 2
P = T + R + S
T = P - S - R
R = t R1 R2
Typically, t ~ 6 – 12 ns
t = coincidence timing window
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

22. A = injected activity T  A, R1  A, R2  A But, R = t R1 R2
Increasing injected activity to compensate for the size of an obese patient not very effective! (randoms rate increases faster than trues rate)
To improve image quality, increase the acquisition time (therefore increasing the signal-to-noise ratio)
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

23. Time window and random coincidence
Coincidence time window: ns
Larger window: higher count rate but more random coincidence counts
More injection activity
 higher single count
rate  more random
coincidence counts c o i n c i d e n c e c i r c u i t

24. Estimating random coincidence
Delay the coincidence time window to acquire pure random count rate
Two coincidence time windows separated by 64 ns (0-12 ns and ns)
No true events in delayed window
No scatter events in delayed window
Randoms count rate same in both windows

25. Major components of a PET scanner
Scintillation detector rings
To convert 511 keV g photons to light photons
PMTs
To convert the light photons to electrons and to greatly multiply the electron number
PMTs being eliminated in some cameras: PET/MR (avalanche photodiodes)
Electronic circuits
To amplify, shape, manipulate and discriminate the electrical signals
Computers
To acquire, process, display, and store images

26. No collimators in a PET scanner
Photon direction determined by LOR  no collimators
Absence of lead improves:
detection efficiency (count rate)
spatial resolution

27. Modes of PET imaging 2-D (with septa):
Coincidences between detectors in the same or nearby ring permitted
3-D (without septa):
Coincidences between detectors in any ring permitted
2-D imaging: high resolution
(reduces randoms + scatter)
3-D imaging: high sensitivity
(increased randoms + scatter)
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

28. PET systems Multiple rings ~5 rings ~8 slices/ring
Ring diameter  80 to 92 cm
Transverse FOV  60 cm
Axial FOV  cm
Attenuation correction
X-ray CT
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

29. 2015. Nuclear Medicine Physics for Radiology Residents
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

30. Detector materials BGO (Bi4Ge3O12) used by GE
LSO (Lu2SiO5) used by Siemens
GSO (Gd2SiO5 ) used by Philips
LYSO (Lu2YSiO5, 9(L):1(Y)) used by all

31. Stopping power (attenuation coefficients)
511 keV keV
NaI /cm /cm
BGO /cm /cm
LSO /cm /cm
GSO /cm /cm
LYSO /cm
_____________________________________________________________________________
N(x) = N e-mx
For x = 3 cm, NBGO = 7.5% (for a 3 cm thick BGO detector, 92.5% of 511 keV photons absorbed)

32. Scintillation decay time
If time interval between detecting 2 g’s too short  pile-up effect  reduced count rate and artifacts
NaI BGO LSO GSO LYSO
230 ns ns ns ns ns

33. NaI BGO LSO GSO LYSO Energy resolution
Depending on both fluctuation of blue photon number and light output
NaI BGO LSO GSO LYSO
10% % 12-18% % %

34. Detector blocks PET
20 – 30 mm
PET spatial resolution is primarily dictated by dimensions of individual detector elements (width ~ 4 – 6 mm)
Each detector element optically isolated by reflective material in crystal cuts
PMT array can determine which detector element(s) absorbed 511 keV photon
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

35. Detector blocks ~300 optically independent blocks in a PET scanner
partial cuts of the detector -> 8 x 8 small crystals
4 PMTs per block.
~20,000 detection elements z R T R T A
detector
optical
coupling
PMT

36. Detector blocks
Each detection element: 4 mm (T) × 4 mm (A) × 30 (R) mm
Small tangential and axial sizes  good spatial resolution
Large radial size  high stopping power  high count rate
All blocks acquire data simultaneously  significantly increasing count rate

37. Assembly of detector blocks
1 block has 8 x = 64 detectors
1 bucket has 4 blocks = 256 detectors
1 ring has 16 buckets = 4096 detectors
5 rings
block 1
block 2
bucket 1
block 3
block 4

38. October 7, Researchers at the University of California, Davis (UC Davis) have received a five-year, $15.5 million grant to develop what they are calling the world's first total-body PET scanner.
National Cancer Institute and will fund the Explorer project, led by Simon Cherry, PhD, distinguished professor of biomedical engineering and Ramsey Badawi, PhD, a professor of radiology.
The total-body PET scanner would image an entire body all at once, and it would acquire images much faster or at a much lower radiation dose by capturing almost all of the available signal from radiopharmaceuticals. … the design would line the entire inside of the PET camera bore with multiple rings of PET detectors.
… such a total-body PET design could reduce radiation dose by a factor of 40 or decrease scanning time from 20 minutes to 30 seconds

39. Advantages of PET imaging
No collimators  higher detection efficiency and better spatial resolution
Ring detectors  higher detection efficiency
Block detectors  higher detection efficiency and better spatial resolution

40. 2014 PET image of the ACR phantom
7.9 mm

41. 2014 SPECT image of the ACR phantomas
9.5 mm
31.8 mm
15.9 mm
Sphere diameters: 9.5, 12.7, 15.9, 19.1, 25.4, and 31.8 mm

42. Time-of-flight PET
Theoretically it is possible to determine the annihilation location from the difference in arrival times of two g photons: d = c∙Dt/2.
Because of fast speed of light (c = 30 cm/ns), fast time resolution of detection is
required for spatial accuracy.
e.g ns  1 cm accuracy
No such fast scintillator yet.
The currently used LYSO for ToF
PET has a time resolution of ns
which leads to 8.8 cm accuracy. d
LOR t1 t2

43. Time-of-flight PET ToF is used to improve SNR.
The improved SNR is used either for better image quality or for shorter scan time.
No additional hardware is needed except more CPU due to the heavy computation.
1 cm t1 t2
0.067 ns
9 cm t1 t2
0.585 ns

44. Time of flight PET image of a big patient
Better resolution with ToF!

45. Iterative reconstruction
© Physics in Nuclear Medicine: Cherry, Sorenson and Phelps

46. LORs combined to form image
Reconstructions - typically OSEM (iterative)
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

47. Iterative reconstruction algorithms
One iteration at one view:
Forward project the image
Compare to acquired data
Backproject P - P0
Update the image
N1 = N0 + bpj (P - P0) N0 P0 P N1
P-P0

48. Iterative reconstruction algorithms
Maximum likelihood - expectation maximization (ML-EM)
accurate but slow convergence
Updates pixel values once after comparisons for all views
Ordered subset - expectation maximization (OSEM)
Updates pixel values after comparison for a subset of views
Number of views in subset increases as convergence occurs
Less iterations needed to achieve same accuracy (faster convergence)

49. Iterative reconstruction algorithms
© Physics in Nuclear Medicine: Cherry, Sorenson and Phelps

50. PET Data Corrections Attenuation (MOST IMPORTANT CORRECTION)
CT based
Normalization
Correction for variation in performance of ~20,000 individual detectors
Random coincidences
Delayed coincidence time window (~64 ns)
Scattered radiation
Modeling from transmission & emission data
Extrapolation from tails of projections
Dead time
Empirical models

51. Photon attenuation within patient
Every PET study is compensated for attenuation.
Correction of attenuation in PET reconstruction needs attenuation map from CT
 values must be extrapolated from CT energies (< 120 keV) to 511 keV
w/o compensated

52. Biograph TruePoint PETCT
Discovery PET/CT 710
(GE)
Biograph TruePoint PETCT
(Siemens)
Ingenuity TF PET/CT
(Philips)
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

53. Attenuation Correction in PET/CT
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

54. Attenuation Correction
Without AC
With AC
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

55. Photon attenuation in PET
High uptake in lungs
Low uptake in others
High uptake in skin
w/o attenuation compensation
attenuation compensation

56. Impact of Misregistration on Attenuation Correction
Lateral walls of myocardium in the PET data, corrected with the lower attenuation of lung tissue
Proper registration
Lateral walls of myocardium corrected with the higher attenuation of heart tissue
(Ref: Attenuation correction of PET cardiac data with low-dose average CT in PET/CT, Tinsu Pan et al, Med. Phys. 33, October 2006)
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

57. Typical PET / CT imaging
(1)
(2)
512 x512 128 x128
120 kVp  511 keV
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

58. Role of FDG (Fluorodeoxyglucose)
18F-FDG is a glucose analog
Actively transported into cells by GLUT (glucose transport proteins)
Both glucose and FDG phosphorylated by hexokinase
Glucose-6-phosphate undergoes further metabolism in the glucose pathway
18F-FDG-6-phosphate does not, and remains trapped in the cell
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

59. Role of FDG
Tumor cells
 increased level of GLUT-1 and GLUT-3  higher levels of hexokinase
 highly metabolically active (high mitotic rates)
 favor the more inefficient anaerobic pathway adding to the already increased glucose demands
These combined mechanisms allow for tumor cells to absorb and retain higher levels of FDG compared to normal tissues
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

60. Role of FDG
NOTE - FDG is NOT cancer specific and will accumulate in areas with high levels of metabolism and glycolysis
 increased uptake can be expected in sites of:
(1) hyperactivity (muscular, nervous)
(2) active inflammation (infection, sarcoid, arthritis, etc.)
(3) tissue repair, etc.
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

61. Hyperactivity
High FDG uptake in pectoralis major after strenuous exercise 24 hours prior to study
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

62. Inadequate fasting 45 min fasting Overnight fasting
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

63. Semiquantitative PET: Standard Uptake Value (SUV)
Defined as the ratio of activity concentrations
SUV = conc. in vol. of tissue / conc. in whole body
SUV = (MBq/kg) / (MBq/kg)
Usually, SUV ~ 2.5 taken as cut-off between malignant and non-malignant pathology
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

64. SUV in clinical studies
Numerator: highest pixel value (SUVmax) from an ROI
Or SUVmean
Denominator: Activity administered/ body mass
Or lean body mass
Or body surface area
SUV will depend on –
physiologic condition, uptake time, fasting state, etc.
Image noise, resolution, ROI definition
Small changes in SUV need to be interpreted carefully
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

65. Requirement for reproducible SUV
18FDG uptake period, scan length, scanning range, scanning direction (e.g. head to toe)
Patient preparation: fasting, medication
Reconstruction parameters: slice thickness, filters
Region-of-interest definition (SUVmax / SUVmean/ body mass/ lean body mass/ body surface area)
Consistency is the most important factor!
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

66. Effect of uptake time
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

67. Clinical Use of PET Oncology (~ 90%) Cardiac & Neuro (~ 10%)
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

68. Typical oncology protocol
Administered dose – 10 to 20 mCi FDG
~ 60 mins. in “quiet” room” to allow adequate uptake and trapping, clearance from blood
Scanning – typically “eyes-to-thighs”
6 to 7 “bed” positions (each ~ 15 cm FOV)
Total scan time is ~ 30 mins. (3 mins/bed)
With time, SUVtumor  & SUVbkg. 
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

69. Patient dose (FDG) Effective dose to pt. Organ of max. dose: bladder
10 mCi (370 MBq) injection  ~ 7 mSv
Organ of max. dose: bladder
Equivalent dose
10 mCi (370 MBq) injection  ~ 63 mGy
CT (for AC) ~ 5 mSv
CT (Diag. / contr.) ~ 15 – 18 mSv
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

70. Personnel dose (FDG) 2.4 Sv / hr per mCi @ 1 m, 1 hr uptake + void
Effective dose (whole body exposure)
10 mCi (370 MBq) injection  ~ 24 1 m
Doses lower towards pt. feet compared to torso
Minimize time of pt. contact at injecting, escorting to rest room, positioning for scan
Increase distance from the pt when communicating
2.4 1 m
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

71. SPECT & PET SPECT – 2 views from opposite sides
Res.  collimator res., which degrades rapidly with increasing distance from collimator face
PET – Simultaneous acquisition
Res.  detector width; is max in center of ring
SPECT sensitivity ~ 0.02%
Huge losses due to absorptive collimators
PET sensitivity- 3D ~ 2% or higher
High sensitivity due to coincidence detection (electronic collimation)
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR

72. Superior spatial resolution Higher sensitivity Attenuation Correction
Advantages of PET over SPECT
Superior spatial resolution
Higher sensitivity
Attenuation Correction
Nuclear Medicine Physics for Radiology Residents Sameer Tipnis, PhD, DABR