6: Positron Emission Tomography What is the principle of PET imaging ? Positron annihilation Electronic collimation – coincidence detection How are the effects of scatter and attenuation corrected for ? What factors can affect resolution ? Examples: PET tracers in oncology and neuroscience After this course you are capable of Describing the essential elements of a PET scan Distinguish the principle of PET detection from that of SPECT Understand the bases of scatter elimination. Understand the factors affecting spatial resolution in PET. Fund BioImag 2013

Positron Emission Tomography (PET) Cancer detection (metastasis) in patients Transaxial Coronal PET/CT Fund BioImag 2013

Positron Emission Tomography Scanner Detector ring Human scanner PET is similar to SPECT but different (detection principle and instrumentation) Fund BioImag 2013

PET of animal models transgenic mice, cancer detection micro PET scanner Positioning of the rodent Imaging gene expression (HSV) Fund BioImag 2013

6-1. What is Positron Emission Tomography ? PET Positron Emission tomography: measured are x-rays emitted by annihilation of positrons emitted by exogenous substance (tracer) in body The principle is as emission tomography, but there is one major difference … (see later) Two issues: How to determine directionality of x-rays ? Absorption is undesirable Injection of positron emitter (tracer) - Most widely used tracer for PET 18Fluoro-deoxy-glucose Fund BioImag 2013

What does one want to measure with PET ? Annihilation photons Question: Why are two photons are produced? Conservation of linear momentum is not possible with one photon (p=E/c) → 2 photons Energie of photons ? hn=mec2=511keV (1eV = 1.6·10-19 J) - + p→n n Unstable parent nucleus p Annihilation coincidence detection : two events detected at same time → annihilation event along a line (defined by detector)  NO need for a collimator NB. Light travels 1m in 3ns : 1[m]/3 108 [m/s]=3ns Fund BioImag 2013

What defines simultaneity (coincidence) ? What is coincidence detection ? electronic collimation (i.e. w/o physical collimators) Electronic signal What defines simultaneity (coincidence) ? Crystal Photomultiplier Leading edge defines time of detection (sharper, i.e. higher 1st derivative) Bi4Ge3O12 (BGO): t~10ns. Position logic electronics Photomultipliers PET-camera Light guide Scintillating crystal Collimator Elimination of collimator material is a major source of sensitivity increase (why?) Fund BioImag 2013

What is really measured with PET ? b b Random coincidences True coincidence a a Scattered coincidence Trues ) Yab = Nab (Aab Tab + Sab + Rab What is measured Scatter Normalization (Instrument imperfection) Fund BioImag 2013 Attenuation Randoms

6-2. Why are Random and Scattered Events bad? mimic a true coincidence emissions from unrelated nuclear transformations interact simultaneously with the detectors Scatter At least one annihilation photon is (Compton) scattered Erroneous Line of incidence (LOI)  assignment to wrong Radon transform Rate of random coincidences : S1 and S2 : count rates on the individual detectors (singles rates)  : coincidence time window. Reduce randoms by reducing t (coincidence interval) Does not work for scattered events (why?) Fund BioImag 2013

How can scattered events be distinguished from true coincidence How can scattered events be distinguished from true coincidence ? Energy discrimination & background subtraction Measure Ef → identify severely scattered photons ENERGY WINDOW ENERGY WINDOW 425 – 650 511 keV photons Counts Energy (keV) 100 200 300 400 700 600 500 Some crystals (BGO) only allow 30% energy discrimination Emission Transmission LSO Scatter BGO Other approaches are needed: Scatter Corrected 350 – 650 Most scattering is by Compton Subtract background (= scatter + randoms) measured in signal void regions → polynomial interpolation ÷ ø ö ç è æ - + = i e f c m E 2 ) cos 1 ( q 2000 Fund BioImag 2013

6-3. How is attenuation correction performed 6-3. How is attenuation correction performed ? simpler for PET than SPECT Probability P1: S1=CT(x)*e-m(d-x)  P1=S1/CT*(x) CT*(x) Probability P2: S2 = CT*(x)e-mx  P2=S2/CT*(x) Attenuation : Probability of detecting the photon pair Compare to geometric average of SPECT (Lesson 5) Fund BioImag 2013 uncorrected corrected

What are the steps in Attenuation Correction for PET ? Mass attenuation coefficient µ/r in soft tissue = 0.095cm2/g (511keV) Average path length for the photon pair longer than for a single photon different lines of response attenuate to varying degrees PET imaging of the body More attenuation (x4) HVL7cm Attenuation correction in practice: Spatially uniform attenuation coefficient assumed Transmission technique using e.g. Cs source (662keV, why is this good enough ?) Less attenuation single photon emitter Comparison with blank scan i.e. subject removed Correction factor for each Radon transform (µ homogeneous) L Fund BioImag 2013

attenuation correction Why is PET/CT the industry standard ? PET-Attenuation correction using CT-Data 0.1 0.2 0.3 m/r (cm/g) CT + PET = PET/CT CT ~70 keV extrapolation Bone PET 511 keV Soft tissue 100 200 300 400 500 Energy (keV) PET/CT: Two separate scanners in one package scatter & attenuation correction CT PET Bone 511 keV CT Fund BioImag 2013 Fund BioImag 2013 6-13 Soft tissue

6-4. Why is Resolution never perfect 6-4. Why is Resolution never perfect ? Annihilation Range and photon non-collinearity Collinearity: Assumed for Reconstruction Background: At time of annihilation, e-p pair has non-zero kinetic energy Mean free path length (range)  conservation of momentum Photon momentum w/zero momentum e-p Range: limits spatial resolution (In air, b+ range ~ several m) pe+pp nucleus + Isotope Half-life (min) Max. Energy (MeV) Range in H2O (FWHM, mm) 18F 110 0.6 1 11C 21 1.0 1.2 15O 2 1.7 1.5 13N 10 1.4 68Ga 68 1.9 82Rb 3.2 e- q ~ 0.25 } ∆x 511 keV 511 keV D (detector distance) D (cm) ∆x (mm) ∆x = 0.5 D tan(0.25) 60 1.3 80 1.7 100 2.2 Fund BioImag 2013

How does the detector affect PET spatial resolution ? Example: BGO Block Detector Coincidence window: 12 ns Energy resolution: ~ 25% Photomultiplier tube True coincidence count rate RT RT=2C*TG2 CT* - tissue activity of a voxel  : the intrinsic detector efficiency (1-e-x) G : the geometric efficiency (solid angle defined by the detector surface/4). NB.  = 0.9→ 81% of photon pairs emitted towards detectors produce coincidence 3 cm scintillator Crystal thickness Resolution is best at the center This is a reason for the 3cm thick crystals used for PET detection. Fund BioImag 2013

6-5. What are typical PET tracers ? Oncology and neuroscience 18Fluoroethyl-Tyrosine (FET) Amino acid transport Deoxy-18fluoro-thymidine (FLT) Proliferation 18Fluoromisonidazole (FMISO) Hypoxia Neuroscience 11C-Methionine Amino acid transport and metabolism H215O Blood flow 18Fluoro-Deoxyglucose (FDG) Glucose metabolism 18FDOPA Presynaptic dopaminergic function 15O-Butanol Blood Flow 11C-Flumazenil Benzodiazepine-receptor mapping FDG or F18 fluorodeoxyglucose O15 Water Fund BioImag 2013

PET: Neuroscience Raw Normal Alzheimer’s Pseudo-color display: red, yellow: high activity green, blue: low activity Fund BioImag 2013

FDG PET Stomach carcinoma Early detection of response to treatment Response-Prediction 3 M. after CTx Day 14 responder Before CTx 6 This is just one of many studies that shows that FDG PET is able to predict response to chemotherapy early after start of the treatment. This is a responding patient with advanced gastric cancer showing increased FDG uptake before treatment and a dramatic decrease of glucose metabolism 2 weeks after the start of treatment. 3 months after chemotherapy, CT verifies the success of neoadjuvant treatment. non-responder Fund BioImag 2013

Whole-body FDG PET Brain Kidney Bladder normal Malignant Lymphoma Fund BioImag 2013

X-ray imaging modalities. Overview CT, SPECT, PET Measurement of signal integrated along line of incidence (LOI) (Radon transform) CT: attenuated incident x-ray beam (direction of beam given by source) SPECT: emitted single photon (need collimation to determine ray direction) PET: annihilation photon pair (directionality by electronic collimation) Apply correction to measured Radon transform (attenuation, scatter, etc.) Backprojection or central slice theorem: Finally an image! CT SPECT PET Projection Encoding Defined by incident x-ray (collimation to reduce scatter) Collimator essential Coincidence detection (electronic collimation) Spatial Resolution (rodent) 100µm-mm (µm) Typical 10mm (Variable and complex) (1.5-3 mm) 4.5-5mm at center (1mm) Attenuation = measurement variable (Varies with energy) Complex correction (Varies with photon energy) Accurate correction (transmission method) Radionuclides None (contrast agents) Any with hn= 60-200keV Positron emitters only Fund BioImag 2013