Positron Emission Tomography Annihilation n The ejected positron

Positron Emission Tomography

Annihilation n The ejected positron (e+) annihilates with an electron (e-) of the tissue after traveling a short distance Positron range Positron Emission Molecule in tissue ee+ Annihilation event 18 FDG The PSF due to positron range has a very long ditail

Radionuclides in PET Isotope Half-life 15 O 2. 0 mins 13 N 10. 0 mins 11 C 20. 5 mins 18 F 1. 8 hrs 81 Rb 4. 58 hrs Energy Positron range 1. 72 Me. V 0. 7 mm 1. 19 Me. V 0. 5 mm 0. 96 Me. V 0. 3 mm 0. 635 Me. V 0. 2 mm

Annihilation Photons n On annihilation, two 511 ke. V annihilation photons are emitted at almost opposite directions Eγ = mec 2 = 511 ke. V 180∘± 0. 25∘ e. Annihilation photon e+ 18 FDG Annihilation photon

Coincidence Detection 180∘± 0. 25∘ detector e e+ Line of response (LOR) 18 FDG t 0 t 1 t 0 -t 1 t 0 t 1

True Coincidences dp ( r ) = a ( r ) 1 2 e - L 1 ( r’) dl e - L 2 ( r’) dl = a ( r ) 1 2 e - L 1 + L 2 ( r’) dl m = a ( r ) 1 2 e - L 1 + L 2 ( r’) dl 1 a ( r) dl L 1+L 2 (Summation) detector L 2 L 1 : As detector efficiency detector 2

Scatter Coincidences n At least one annihilation photon experiences (Compton) scattering before detection True LOR Assigned LOR

Scatter Coincidences (Cont’s) n Contribute to a low-resolution background ü Degraded image contrast and spatial resolution n Depend on: ü Attenuation distribution ü System configuration ü energy resolution of the detector • BGO detector 350 ke. V scatter window : 10 -15% scatter fraction n Difficult to correct

Random Coincidences n Simutaneous detection, as defined by the coincidence window, of two uncorrelated annihilation photons by change Assigned LOR

Random Coincidences (Cont’s) n Contribute to a smooth background ü estimated from single rates: R = 2 C 1 C 2 Detector 2 ü using delayed lines Detector 1 n Depend on: ü Coincidence resolving time ü Detector material delay Coincidence detection n Correction possible:

Detector Design Four 1” Square PMT Light guide BGO PMT 8× 8 Individual coupling Block detector BGO crystal block Sawed inot 64 seg. , each 6 mm square

Deposited Energy Photo-peak or full-energy peak d. N/d. E Compton continuum ( ) Compton edge ( ) 511 ke. V E

Energy Resolution Counts Photopeak 350 511 Energy (ke. V)

Physical Factors Affecting Resolution 1. Positron range FWHM = s, related to isotope 2. Photon noncolinearity FWHM = 0. 0022 D, D = ring diameter 3. Detector width, d 4. Intercrystal scatter 5. Light sharing + positron logic FWHM 1. 25 √(d/2)2 +s 2 + (0. 0022 D) 2 + b 2 b 2 mm for block detectors, 0 mm for individual coupled detectors JNM, 34, 101, 1993

Scintillation Crystal n detection efficiency for 511 ke. V photons ü effective Z number ü density ü photoelectric effect preferred n light yield ü energy resolution ü scatter rejection n scintillation decay constant ü timing resolution ü reducing randoms ü TOF (time-of flight) PET n cost, mechanic properties, refractive index, etc.

Scintillation Crystal 4 Stopping power: – Effective atomic number (Iodine: 53, relatively high) – Density: 3. 76 g/cm 3 4 Light yield: 38 photons/ke. V (4 e. V/per photon) – – Good light yield, used as reference = 100 Energy resolution (Poisson statics) no. generated proportional to deposited energy 15% scintillation Efficiency 4 Light decay constant: 230 s after glow – Dead time – Position mis-positioning – Wavelength at max. emission: 415 nm 4 Reflective index: 1. 85 – Hygroscopic, relatively fragile

Common Inorganic Crystals Scintillator Wave length (nm) Decay constant (ns) Refraction index Density (g/cm 3) Light yield Na. I (Tl) 410 230 1. 85 3. 67 100 Cs. I (Na) 420 630 1. 84 4. 51 85 Cs. I (Tl) 565 1000 1. 80 4. 51 45 Li. I (Eu) 470 -485 1400 1. 96 4. 08 35 Ca. F 4 435 900 1. 44 3. 19 50 BGO 480 300 2. 15 7. 13 15 GSO 410 60 1. 9 6. 71 16 Ba. F 2 225/310 0. 6/620 1. 49 4. 89 4/20 Cd. WO 4 540 5 2. 2 7. 9 40 LSO 480 40 1. 82 7. 4 75 YSO 420 70 1. 80 4. 54 118 ?

Crystal vs. Light yield Na. I (Tl) Light yield Cs. I (Tl) Cs. I (Na) 420 410 565 Wavelength (nm)

Parallax Errors 4 Point source 下，detector width 小、detector length 長，則 resolution 好。 4 Eccentric point source 下，則會有 radial & tangential projections 影響 resolution。radial 投射尤其不利。 Positron source Radial projection Tangential projection

Parallax Errors

Noise Equivalent Count (NEC) 3. 0 105 2. 5 105 NEC 2. 0 105 1. 5 105 1. 0 105 NEC = T 2 / (T+S+R) 5. 0 104 0 0 0. 2 0. 4 0. 6 0. 8 1. 0 Activity [u. Ci/ml] 1. 2 1. 4

PET Imaging Model Continuous Model: R = Rt + Rs + R a Rt = α a(x, y, z) h(x, y, z) dx dy dz y = R t + n Discrete Approximation: y = Ha + n α: attenuation n : Poison noise H : system geometry y : known data a : unknown data

PET Image Reconstruction 4 1. Filtered Backprojection (FBP) 4 2. Iterative Methods – EM, OSEM – Bayesian – PWLS

System Configurations Multi-ring systems : usually BGO crystal PENN PET : Na. I crystal in 6 heads ü C PET : Na. I crystal in 2 -3 curved heads 3. Time-of flight PET : (TOF PET) : Ba. F 2 crystal ü ultra-short decay constant 4. SPECT for coincidence imaging 1. 2.

What Use PET (from LPP) 1. The basis of all tissue function is chemical. 2. Diseases result from errors introduced into chemical by viruses, bacteria, genetic abnormalities, drugs, environmental factors, aging and behavior. 3. The most selective, specific, and appropriate therapy is one chosen from a diagnostic measure of the basic chemical abnormality. 4. Detection of chemical abnormalities provides the earliest identification of disease, even in a presymptomatic stages before the disease process has exhausted the chemical reserves or over-ridden the compensatory mechanisms of the brain. 5. Assessment of restoration of chemical function provides

SPECT reconstruction: 4 Issues: attenuation, scatter, noise, DDSR, sampling geometry 4 Filtered Backprojection (FBP) – ignore attenuation, DDSR – usually no scatter correction – ad hoc smoothing for controlling image noise 4 Iterative Reconstruction – OSEM – allow attenuation, and DDSR corrections – optimal noise control – usually no scatter correction – needs attenuation map 4 Analytical approaches uniform attenuation 4 Simultaneous Emission, Attenuation map Reconstruction 4 Dynamic SPECT by interpolation vs. timing

Advantage of PET over SPECT 1. Simple model 1. attenuation 2. not DDSR (distance dependent systemic resolution) 2. Higher sensitivity 1. electronic collimation 2. large solid angle of detection 3. Better tracers availability

New Trends (3 D PET) n remove septa to allow coincidence detection between detectors at different rings n advantage: ü increase detection sensitivity n performance issue: ü increased scatter ü increased randoms ü parallax errors in axial direction n reconstruction: ü FORE (Fourier Rebinning) ü 3 D iterative methods ü hybrid

New Trends (DOI Detectors) Depth-of-Interaction (DOI) Detectors n detectors capable of providing depth information n used for reducing parallax errors n increased sensitivity n reduced ringer diameter n current depth resolution: 5~10 mm

New Trends (DOI Detectors) 4 Images of a point source displaced 10 cm from the center with 3 mm × 30 mm crystals 5 mm DOI resolution no DOI information

New Trends (DOI Detector 1) 1” Square Photomultiplier Tube Array of 64 photodetectors 1” 64 BGO crystals each 3 mm square Block detector 30 mm 1”

New Trends (DOI Detector 2) LSO GSO or LSO PMT pulse shape discrimination circuit decay constant : depends on crystal 中雜質 Phoswich detector ( by CTI System )

New Trends (DOI Detector 3)

New Trends (TOF Detector) n Time-of-Flight PET Block detector employing light sharing 2 D Wire Chamber Readout 2” Block detector Gamma to Electron Converter 2” 8 -16 mm Ba. F 2/TMAE or Metal (Pb or W) Foils

New Trends (DOI PET system) n Siemens/CTI HERRT System 31. 2 cm 46. 9 cm Crystal length: 7. 5 mm × 2

New Trends (DOI PET System) LSO (7. 5 mm) GSO (7. 5 mm) light guide PMT crystal block Rows 19. 5 mm Columns

New Trends (Small-Animal Systems) n dedicated for small-animal studies ügene expression, gene transfer üdrug effects übasic physiology, etc. n requirements: ühigh resolution ühigh sensitivity n configuration: üsmall diameter üelongated detectors

New Trends (Small-Animal Systems)

New Trends (gene expression)

New Trends (gene expression)

New Trends (gene expression)

New Trends (drug effects)

New Trends (basic physiology)

New Trends (Small-Animal System 1) 2 mm × 10 mm LSO 17. 2 cm diameter detector ring 110 mm transaxial FOV, 18 mm axial FOV UCLA

Volume resolution in mm New Trends (Small-Animal System 1) 64 mm 3 20 Rx × Ry × Rz 15 Rz 10 Ry Rx 8 mm 3 5 0 0 10 20 30 Offset in mm 40 50 micro. PET EXACT HR+ UCLA

New Trends (Small-Animal System 1) 20 Phantom size vs Scatter cps 15 NEC Medium (350 -650 ke. V) (× 105) NEC Large (350 -650 ke. V) 10 NEC Small (350 -650 ke. V) 5 0 0 5 10 15 Activity (u. Ci/cc) 20 25

New Trends (Small-Animal System 2) Germany

New Trends (Small-Animal System 2) Tier PET 4 YAP scintillator 4 2 mm × 15 mm crystal 4 variable detector-to detector distance 4 transaxial FOV: 40 mm 4 resolution: 2. 1 mm Germany

New Trends (Compact Systems) n systems capable utilizing the entire space inside the detector ring for imaging üCurrent systems utilization is about 60 -70% n advantages: üimproved sensitivity üimproved resolution üreduced cost n disadvantages: üincreased scatter üincreased randoms

New Trends (Compact System 1) FOV 85. 9 mm 57. 3 mm 56. 3 mm RFOV = 56. 3 mm Compact Conventional

New Trends (Compact System 1)

ML-EM results True REBIN+FBP SREM+FBP Conventional Ring 57. 3 mm 89. 5 mm RFOV 56. 3 mm

New Trends (Compact System 2) UCLA

New Trends (Hybrid systems) n Systems capable of providing co-registered multimodal imaging n Providing complementary diagnostic information

New Trends (Hybrid system) n Combined PET / CT scanner

New Trends (Hybrid system) n Combined PET / CT scanner + Function (PET) = Anatomy (CT) Fusion (PET/CT)

New Trends (Hybrid systems) n PET / CT ücrystal for CT, crystal for PET üeven Adds DOI detector n PET / MRI üPMT sensitive to magnetic field n PET / SPECT n Novel phoswich detector design for multimodal imaging

New Trends (Miscellaneous) n list-mode data acquisition ü Combined 3 D & DOI PET : LOR as millions or more, requires more storage spaces or list-mode acquisition Activity n spatial-temporal reconstruction ü Poor S/N ratio in later frames in real image acquisition Real Ideal time

New Trends (laboratory PET)