Lecture-5-gamma-cameras.pdf
- 1. Imaging and detectors for medical physics Dr Barbara Camanzi barbara.camanzi@stfc.ac.uk Lecture 5: Gamma cameras Joint CI-JAI advanced accelerator lecture series
- 2. Course layout Day AM 09.30 – 11.00 PM 15.30 – 17.00 Week 1 6th June Lecture 1: Introduction to medical imaging Lecture 2: Detectors for medical imaging 7th June Lecture 3: X-ray imaging 8th June Tutorial Week 2 13th June Lecture 4: Radionuclides 14th June Lecture 5: Gamma cameras Lecture 6: SPECT 16th June Lecture 7: PET Week 3 22nd June Tutorial Page 2/42
- 3. Books Page 3/42 1. N Barrie Smith & A Webb Introduction to Medical Imaging Cambridge University Press 2. Edited by M A Flower Webb’s Physics of Medical Imaging CRC Press 3. A Del Guerra Ionizing Radiation Detectors for Medical Imaging World Scientific 4. W R Leo Techniques for Nuclear and Particle Physics Experiments Springer-Verlag
- 4. The gamma camera Ref. 1 – Chapter 3.6 • Gamma camera= instrumental basis for: 1. SPECT 2. Planar scintigraphy • → used in nuclear medicine imaging = emission imaging = radioactive source inside patient’s body • In both SPECT and planar scintigraphy radiation source used = mainly 𝑇𝑐𝑚 99 → 140 keV 𝛾-rays → hence the name gamma camera • Patient lies beneath gamma camera positioned close to organ of interest Page 4/42
- 5. Gamma camera operation • Detects 𝛾-rays and for each 𝛾-ray records: 1. x-y position 2. Energy 3. Time → can be linked to other information (for ex. ECG) • Main features: 1. Capable of withstanding 𝛾-ray detection rates up to tens to thousands events per second 2. Reject 𝛾-rays scattered in the body → no useful spatial information 3. As high sensitivity as possible → high quality images within clinically acceptable imaging time Page 5/42
- 6. Page 6/42 Gamma camera components and operation • Gamma camera components: 1. Collimator 2. Scintillator 3. Array of PMTs 4. ‘Anger’ position network 5. Pulse height analyser (PHA) 6. Digitizer = ADCs 7. Computer to build up the image from many detection events • Operation: 1. 𝛾-rays from patient travelling in certain directions selected 2. 𝛾-rays produce scintillation 3. Scintillation detected by PMTs ‘Anger’ position network PHA Digitizer Computer
- 7. Collimator • 𝛾-rays from radioactive source within the body emitted in all directions → high degree of collimation needed • Operating principle: absorptive collimation Only 𝛾-rays travelling in certain directions go through and reach the detector, the others are absorbed in the septa → most radiation is absorbed → inefficient • Made usually from tungsten or lead to provide high absorption probability Page 7/42
- 8. Page 8/42 • Array of holes drilled or cast in tungsten / lead or shaped from tungsten / lead foils • Hexagonally-based ‘honeycomb’ geometry • Septal thickness chosen to maximise attenuation (95%) and efficiency (sensitivity) Collimator design
- 9. Page 9/42 Collimator basic designs • Six basic designs: 1. Parallel-hole: most common design 2. Converging 3. Diverging 4. Pinhole 5. Slanthole 6. Fan-beam Parallel-hole Diverging Slanthole Converging Pinhole
- 10. Page 10/42 Parallel-hole collimator 𝐼 = image object size 𝑂 = object size Magnification 𝑀 = 𝐼 𝑂 = 1 • Holes are all parallel to each other • Provides not magnified, not inverted images • Comes in a variety of forms in which hole size, length and septal thickness are traded off to match energy of 𝛾-ray Courtesy Piero Posocco (Imperial College) Pinholes Detector Collimator 𝑶 𝑰
- 11. Page 11/42 Parallel-hole collimator resolution • Resolution of collimator 𝛿𝑟𝑐𝑜𝑙𝑙 is given by: 𝛿𝑟𝑐𝑜𝑙𝑙 ≈ 𝑑 𝑙 + 𝑧 𝑙 𝑑 = hole size 𝑙 =septal length 𝑧 =distance object – collimator = depth within the body • Resolution: 1. Independent of septal thickness 𝑡 2. Gets worst with 𝑧 = depth in body and camera distance from body Object 𝑂 Image 𝐼 𝑡 𝑑 𝛿𝑟𝑐𝑜𝑙𝑙 Courtesy Piero Posocco (Imperial College) Septal thickness
- 12. Page 12/42 Parallel-hole collimator efficiency • Geometric efficiency 𝐺 defined as fraction of 𝛾-rays detected = transmitted by collimator divided by total number of 𝛾-rays emitted • Given by: 𝐺 ≈ 𝑑4 12𝑙2 𝑑 + 𝑡 2 • Assumption: attenuation in body ignored • Efficiency independent of 𝑧 𝑑 = hole size 𝑡 = septa thickness 𝑙 = septa thickness 𝑧 = distance sourced - collimator Transmitted rays 𝑙 𝑧 𝑑 𝑡 Point source emitting over 4𝜋 solid angle Courtesy Piero Posocco (Imperial College)
- 13. Parallel-hole collimator: resolution –vs– efficiency • Bigger hole size → worst spatial resolution but higher efficiency → trade-off between the two: • Example: Page 13/42 Parameters High resolution High efficiency Septa thickness 𝑡 (mm) ~0.2 ~0.2 Septal length 𝑙 (mm) 24 24 Hole size 𝑑 (mm) 2 4 Spatial resolution 𝛿𝑟𝑐𝑜𝑙𝑙 (mm) 1 10 21 Efficiency 𝐺 (%) 4.8 × 10−4 2.1 × 10−3 1For 𝑧 = 100 mm 𝛿𝑟𝑐𝑜𝑙𝑙 ≈ 𝑑 𝑙 + 𝑧 𝑙 𝐺 = 𝑑4 12𝑙2 𝑑 + 𝑡 2
- 14. Use of parallel-hole collimators with radioisotopes • Used with 𝑇𝑐𝑚 99 (𝐸𝛾 = 140 keV) called low-energy collimators: 1. Smaller holes → called low-energy high-resolution (LEHR) collimators 2. Large holes → called low-energy all-purpose (LEAP) collimators • Used with 𝐺𝑎/ 𝐼 111 11 and 𝐼 131 called medium-, high-energy collimators: 1. Septa of increasing thickness Page 14/42
- 15. Page 15/42 Converging collimator 𝐼/𝑂 = image / object size 𝑙 = collimator thickness 𝑧 = distance object – collimator, 𝑓 = focal distance Magnification 𝑀 = 𝐼 𝑂 = 𝑓+𝑙 𝑓−𝑧 • Holes converge to a point in front of collimator = focused toward body • Provides magnified images • Focal point typically 40÷50 cm in front of collimator: 1. For objects between collimator and convergence point → magnified non-inverted image 2. beyond convergence point → magnified inverted image • Centre of curvature ideally in the middle of imaging FOV Detector Collimator 𝑓 𝑙 𝑧 𝑶 𝑰 Courtesy Piero Posocco (Imperial College)
- 16. Page 16/42 𝐼, 𝑂, 𝑙, 𝑧, 𝑓 = as in previous slide De-magnification 𝑀 = 𝐼 𝑂 = 𝑓−𝑙 𝑓+𝑧 Detector Collimator Diverging collimator • Converging collimator turned around → holes diverge from detector face = focused away from body • Focal point typically 40÷50 cm behind collimator • Provides de-magnified non-inverted images of size dependent on distance → distortions • Larger FOV than parallel hole collimator • Use: whole-body planar scintigraphy 𝑶 𝑰 𝑙 𝑓 𝑧 Courtesy Piero Posocco (Imperial College)
- 17. Page 17/42 Detector Collimator 𝐼/𝑂 = image / object size 𝑧 = distance object – collimator, 𝑓 = focal distance De-magnification 𝑀 = 𝐼 𝑂 = 𝑓 𝑧 Pinhole collimator • Single hole with interchangeable inserts of 3,4,6 mm aperture in led, tungsten, gold, etc. • Provides inverted images with significant (de)-magnification and high spatial resolution at small 𝑧 but also geometric distortion at image edges due to magnification changing with distance: 1. For 𝑧 < 𝑓 image magnified 2. For 𝑧 > 𝑓 image de-magnified • Use: image very small organs such as thyroid and parathyroid 𝑶 𝑰 𝑓 𝑧 Courtesy Piero Posocco (Imperial College)
- 18. Page 18/42 Slanthole collimator • Parallel septa all tilted at same angle with respect to detection scintillator crystals • Use: primarily in breast and cardiac imaging
- 19. Fan-beam collimator • Two different geometries in the two directions: 1. Head/foot direction = holes are parallel → parallel-hole collimator 2. Radial direction = holes are converging → converging collimator • Provides image magnification over reduced FOV • Use: primarily for brain and heart studies Page 19/42
- 20. Page 20/42 Scintillator • Scintillation detector = large single 𝑁𝑎𝐼 𝑇𝑙 crystal of dimensions: 1. Diameter = 40÷50 cm 2. Thickness = 6÷12.5 cm chosen as compromise = the thicker the crystal: a. Broader LSF → lower spatial resolution b. Higher number of 𝛾-rays detected → higher 𝑆𝑁𝑅 • 𝑁𝑎𝐼 𝑇𝑙 hygroscopic → needs hermetic sealing: 1. Aluminium outside 2. Glass toward PMT
- 21. 𝑵𝒂𝑰 𝑻𝒍 characteristics Parameter Value Linear attenuation coefficient at 140 keV 2.2 cm−1 Decay time 230 ns Peak emission wavelength 415 nm Photon yield 38 𝛾/keV Page 21/42 • Transparent to 415 nm light → little or no energy lost due to reabsorption • Emission wavelength (415 nm) produced well-matched to optimal performance of conventional PMTs • Photon yield (number of photons emitted) = directly proportional to 𝐸𝛾 • Large crystals grown easily → cheap
- 22. Page 22/42 PMT array • Standard PMTs 2÷3 cm in diameter → array of 61, 75 or 91 PMTs used to cover the large crystal area • Most efficient geometry: hexagonally-close pack → same distance from centre of one PMT to centre of each neighbouring PMT → important to reconstruct position of scintillation event • Thin optical coupling layer
- 23. PMT calibration • If PMT response across array not uniform → image artefacts • Level of non-uniformity tolerated: a. Up to 10% in planar scans b. < 1% in SPECT • PMT calibrated using sample of known radioactivity → calibration coefficient calculated and applied to the data – New calibration systems use as calibration source one LED per PMT → continuous calibration also during scans Page 23/42
- 24. ‘Anger’ position network • Scintillation event in 𝑁𝑎𝐼 𝑇𝑙 → signal in PMTs: – PMT closest to event → largest output signal – Adjacent PMTs → smaller output signals = light detected ~ inversely proportional to distance event – PMT • Position of scintillation event calculated by comparing magnitudes of PMTs output signals = centroid or centre of mass of light distribution in PMTs • More complex positioning algorithms that account for local spatial distortions and apply corrections for PMT non-uniform performance can also be used Page 24/42
- 25. Page 25/42 ‘Anger’ position network something extra • In simplest (linear) case the position is given by: 𝐷1 = 𝑆2 × 𝐷 𝑆1 + 𝑆2 𝐷2 = 𝑆1 × 𝐷 𝑆1 + 𝑆2 𝐷1, 𝐷2 = distances scintillation event – centre of PMT 𝐷 = 𝐷1 + 𝐷2 𝑆1, 𝑆2 = PMT output signals Courtesy Piero Posocco (Imperial College)
- 26. Pulse height analyser • The pulse height analyser determines the recorded events where: 1. The 𝛾-ray has not been scattered in the patient = good event → to be retained 2. The 𝛾-ray has been Compton scattered in the patient → lost spatial information = bad event → to be rejected • The pulse height analyser allows spectroscopic imaging = selects for display only events falling within a particular energy range Page 26/42
- 27. Pulse height analyser operating principle • Amplitude of PMT output signal proportional to 𝐸𝛾: 1. Amplitude of PMT output signal proportional to number of optical photons arriving from scintillator 2. Number of optical photons produced in scintillator proportional to 𝐸𝛾 • Discriminating on the basis of amplitude of PMT output signal = discriminating on the basis of 𝐸𝛾 • The pulse height analyser takes summed electrical signals from the PMT array and applies thresholds Page 27/42
- 28. Pulse height analyser operation • Two steps: 1. The PMT output signals are sent to the multiple-channel analyser → pulse-height histogram 2. Two voltage thresholds, upper and lower, are applied to the pulse-height histogram to select the 𝛾-rays to be accepted Page 28/42
- 29. Page 29/42 Pulse height analyser multiple-channel analyser • Multiple-channel analyser (MCA): 1. Receives in input PMT output signals 2. ADC digitizes the signals 3. Produces the pulse-height histogram = plot of number of events from the PMTs as function of output amplitude • Channel = specific energy range → number of channels can be > 1000 → complete energy spectra produced Courtesy Piero Posocco (Imperial College)
- 30. Pulse height analyser thresholds • Variation in 𝐸𝛾 detected for various reasons: 1. Non-uniformity in PMT and scintillator response → range of outputs for mono-energetic 𝛾-rays 2. Compton scattering of 𝛾-rays → variable lower 𝐸𝛾: a. Scattering in patient at small angles still has useful spatial information → event needs to be accepted b. Scattering in scintillator retains useful spatial information → event needs to be accepted • One voltage level at photopeak = full 𝐸𝛾 to select accepted event not enough → energy range = detection window needed → two thresholds Page 30/42
- 31. Page 31/42 Detection window in context • Isomeric transition of 𝑇𝑐𝑚 99 → 140 keV 𝛾-ray • Peaks in pulse-height diagram: – A = total absorption of 140 keV 𝛾-ray by scintillator – B = Iodine escape peak = 𝛾-ray excites iodine → K-shell X-ray emitted → escapes scintillator → detected 𝐸𝛾 28÷33 keV lower – C = photoelectric absorption in lead collimator → K-shell X-rays at ~78 keV → absorbed in scintillator • Detection window set to accept A and B but reject C Courtesy Piero Posocco (Imperial College)
- 32. Page 32/42 Types of events • A = good event • B = scatter in detector – 𝛾-ray scatters in detector → position information distorted – Full energy deposited → accepted • C = scatter in patient – 𝛾-ray scatters in patient and may retain sufficient energy to fall in energy window – unwanted event • D = septal penetration – 𝛾-ray penetrates through wall of collimator and reaches detectors – unwanted event Patient with 𝛾-emitting radionuclide concentrated in liver Courtesy Piero Posocco (Imperial College) Collimator Light guide PMTs Scintillator
- 33. Gamma camera energy resolution • Energy resolution defined as FWHM of photopeak → the narrower FWHM the better is the gamma camera at discriminating unscattered and scattered 𝛾-rays • Typical energy resolution values: 1. Without patient: 14 keV = 10% → measured during calibration 2. With patient: 28 keV = 20% → set at this value to increase the acceptance window Ex.: 20% window around photopeak = 20% around 140 keV → 𝛾-rays in energy range 127÷153 keV accepted Page 33/42
- 34. Gamma camera detection efficiency • Gamma camera detection efficiency = fraction of emitted 𝛾-rays that produce counts in the image • The detection efficiency Ε𝑐𝑎𝑚𝑒𝑟𝑎 made by three terms: Ε𝑐𝑎𝑚𝑒𝑟𝑎 = 𝐺 ∙ 𝜀 ∙ 𝑓 𝐺 = collimator contribution 𝜀 = scintillator contribution 𝑓 = read-out contribution Page 34/42
- 35. Terms contributing to the detection efficiency 1. Collimator → geometric collimator efficiency 2. Scintillator → ratio of 𝛾-rays recorded by the scintillator 𝑁𝑟𝑒𝑐𝑜𝑟𝑑 to the 𝛾-rays hitting it 𝑁ℎ𝑖𝑡: 𝜀 = 𝑁𝑟𝑒𝑐𝑜𝑟𝑑 𝑁ℎ𝑖𝑡 = 1 − exp −𝜇 ∙ 𝑑 𝜇 = scintillator linear attenuation coefficient at 𝐸𝛾 𝑑 = scintillator thickness 3. Read-out → fraction of photons accepted by energy discriminating unit Page 35/42
- 36. Dead-time • If injected dose very large1 number of 𝛾-rays arriving on scintillator can exceed system recording capabilities due to electronics finite recovery and reset times → events get lost during the time the system cannot respond = cannot record the events → dead-time • Two types of behaviour: 1. Paralysable 2. Non-paralysable • Dead-time determined by non-paralysable behaviour 1This is particularly true at start of scan Page 36/42
- 37. Paralysable / non-paralysable behaviour Paralysable behaviour Component cannot respond to new event until after a certain time from previous event Example: Each time a 𝛾-ray the scintillator produces excited state that decays after 230 ns. If another 𝛾-ray strike the scintillator on the same spot before the excite state decay, it will take another 230 ns for the state to decay and only one set of photons will be produced. The higher the rate of 𝛾-rays striking the detector, the longer the dead-time Page 37/42 Non-paralysable behaviour Component cannot respond for a set time irrespective of level of radioactivity Example: The PHA takes a certain time to process the input signal from an event. During this time further events are not recorded independently from how many they are
- 38. Determination of dead-time • The dead-time 𝜏 is given by: 𝜏 = 𝑁 − 𝑛 𝑛𝑁 𝑁 = true count rate = number of scintillations / s 𝑛 = measure count rate • Maximum measurable count rate = 1 𝜏 • Dead-time corrections based on calibration using phantoms of known radioactivity Page 38/42
- 39. Gamma camera spatial resolution • Total gamma camera spatial resolution 𝛿𝑟𝑡𝑜𝑡 given by sum of three terms: 𝛿𝑟𝑡𝑜𝑡 = 𝛿𝑟𝑐𝑜𝑙𝑙 2 + 𝛿𝑟𝑖𝑛𝑡 2 + 𝛿𝑟𝐶𝑜𝑚𝑝𝑡𝑜𝑛 2 𝛿𝑟𝑐𝑜𝑙𝑙 = collimator spatial resolution 𝛿𝑟𝑖𝑛𝑡 = gamma camera intrinsic resolution 𝛿𝑟𝐶𝑜𝑚𝑝𝑡𝑜𝑛 = Compton scattering contribution • Typical values for overall spatial resolution are: 1. 1÷2 cm deep in the body 2. 5÷8 cm close to surface of body → close to collimator Page 39/42
- 40. Terms contributing to spatial resolution 1. Collimator spatial resolution: a. Determined by: i. Collimator geometry ii. Depth of source = organ where radiotracer is accumulated b. Typical values 5÷8 mm c. Typically dominant term to total spatial resolution 2. Gamma camera intrinsic resolution a. Due to: i. Uncertainty in exact localisation of point where scintillation occurs = ability of PMTs to localise events ii. Intrinsic resolution of positioning system b. Typical values 3÷5 mm 2. Compton scattering contribution a. Due to 𝛾-rays undergoing Compton scattering in the body Page 40/42
- 41. Page 41/42 Signal-to-noise ratio • Radioactive decay = statistical process → number of disintegrations per unit time fluctuates • Poisson distribution → 𝑆𝑁𝑅 depends on number of detected 𝛾-rays 𝑁: 𝑆𝑁𝑅 ∝ 𝑁 • Factors affecting 𝑆𝑁𝑅: 1. Administered radioactivity 2. Radiotracer uptake 3. Depth of organ in body = the deeper the organ the more attenuation → lower 𝑆𝑁𝑅 4. Total scanning time 5. Intrinsic resolution of the gamma camera → only collimator properties can be changed 6. Post-acquisition filtering → increases 𝑆𝑁𝑅 but adds blur
- 42. Contrast and contrast-to-noise ratio • Theoretically no noise → contrast and 𝐶𝑁𝑅 very high No background from tissues of no interest = where radiotracer does not accumulated • In reality Compton scattered 𝛾-rays → random background signal higher close to region with uptake of radiotracer Page 42/42