Nuclear medicine imaging: PET and SPECT

In short

Nuclear medicine imaging maps function rather than anatomy: a radiopharmaceutical is given to the patient, it concentrates where a physiological process is active, and the gamma rays it emits are detected outside the body to form an image. SPECT (single photon emission computed tomography) detects single gamma photons (typically 140 keV from Tc-99m) through a mechanical collimator; PET (positron emission tomography) detects the two back-to-back 511 keV photons from positron annihilation in electronic coincidence, so it needs no collimator and is far more sensitive and quantitative. The topic is examined in Medical Imaging, Radiation Physics and Biomedical Instrumentation papers, in GATE Biomedical Engineering (BM) under "nuclear imaging / radiation detectors", and in AMIE/university B.E. papers. In India it underpins work at Tata Memorial/ACTREC, AIIMS, PGIMER, SGPGI, HCG, Apollo and Fortis nuclear medicine departments, at BRIT (Board of Radiation and Isotope Technology) and Jubilant Radiopharma, and in service/application roles with GE Healthcare, Siemens Healthineers and Philips. All such facilities are licensed and regulated by the AERB under the Atomic Energy Act, 1962.

Explained

Physics of the signal

• Radioactive decay: activity A = λN decays exponentially; unit becquerel (Bq) = 1 disintegration/s. The old unit curie: 1 Ci = 3.7 × 10¹⁰ Bq, 1 mCi = 37 MBq.
• Gamma emission / isomeric transition: Tc-99m decays to Tc-99 emitting a single 140.5 keV photon - nearly ideal (penetrates tissue, stopped efficiently by a thin NaI crystal, no particulate dose).
• Positron (β⁺) decay: a proton converts to a neutron, emitting a positron and a neutrino. The positron travels a short distance (positron range, ~0.6 mm in water for F-18, several mm for Rb-82), thermalises and annihilates with an electron.
• Annihilation: rest mass converts to two photons of 511 keV each (E = m₀c², m₀c² = 511 keV per electron), emitted ~180° apart. The two unavoidable physical blurs are positron range and non-collinearity (±0.25°, giving ~0.0022 D blur for ring diameter D).
• Interaction in tissue: at 140 keV Compton scatter dominates in soft tissue (μ ≈ 0.15 cm⁻¹ in water); at 511 keV μ ≈ 0.096 cm⁻¹. Hence attenuation and scatter correction are essential for quantification.
• Counting statistics: emission is Poisson, so noise SD = √N and SNR = √N. Halving the dose or time worsens SNR by √2 - the fundamental dose-versus-noise trade-off.

Radionuclides and radiopharmaceuticals

• Production routes: *reactor* (fission/neutron activation - Mo-99, I-131, Lu-177), *cyclotron* (proton-rich, β⁺ emitters - F-18, C-11, N-13, O-15), *generator* (Mo-99/Tc-99m "cow", Ge-68/Ga-68).
• Mo-99/Tc-99m generator: Mo-99 (T½ 66 h) adsorbed on alumina decays to Tc-99m; sterile saline elutes Tc-99m-pertechnetate daily. This is why Tc-99m is used in ~80% of all SPECT studies worldwide.
• Common SPECT agents: Tc-99m MDP (bone), Tc-99m sestamibi/tetrofosmin (myocardial perfusion, parathyroid), Tc-99m DTPA/EC/MAG3 (renal), Tc-99m MAA (lung perfusion), Tc-99m pertechnetate (thyroid), Tc-99m sulphur colloid/nanocolloid (sentinel node), I-123/I-131 (thyroid), Tl-201 (cardiac, older), In-111/Ga-67 (infection, older).
• Common PET agents: F-18 FDG (fluorodeoxyglucose - glucose analogue trapped as FDG-6-phosphate; oncology, neurology, cardiac viability, infection), F-18 NaF (bone), Ga-68/F-18 PSMA (prostate cancer), Ga-68 DOTATATE (neuroendocrine tumours), C-11 choline/methionine, N-13 ammonia and O-15 water (perfusion), F-18 DOPA, amyloid and tau tracers.
• Half-lives (physical): O-15 2.04 min; N-13 9.97 min; C-11 20.4 min; F-18 109.7 min; Ga-68 67.7 min; Tc-99m 6.01 h; I-123 13.2 h; In-111 2.8 d; I-131 8.02 d; Lu-177 6.65 d.
• Theranostics: the same targeting molecule labelled with an imaging nuclide (Ga-68 DOTATATE / PSMA) and a therapy nuclide (Lu-177, I-131, Ac-225) - imaging selects the patient, therapy treats them. Widely practised in Indian tertiary centres.
• Logistics: F-18 needs a cyclotron within a few hours' transport; Ga-68 and Tc-99m come from generators on site. India has medical cyclotrons at BARC/RMC (Parel), Tata Memorial/ACTREC, AIIMS, SGPGI, PGIMER and several private networks; BRIT distributes reactor isotopes from Navi Mumbai.

SPECT: the gamma camera and tomography

Anger (scintillation) camera chain - in order:

1. Collimator - a lead plate with thousands of holes that accepts only photons travelling nearly perpendicular to the crystal. It defines direction mechanically and rejects >99.9% of photons, which is why SPECT sensitivity is low (~10⁻⁴).
   • Types: parallel-hole (most common; constant size, FOV = detector size), converging (magnifies, small organs), diverging (minifies), pinhole (high resolution for thyroid, small animal).
   • Grades by energy/resolution: LEHR (low-energy high-resolution), LEAP/LEGP (general purpose), ME and HE (medium/high energy for I-131, Ga-67). Resolution and sensitivity trade off against each other.
2. Scintillator - NaI(Tl) crystal, typically 9.5 mm (3/8 in) thick, 40-55 cm across; high light output, hygroscopic, needs hermetic sealing. Converts a gamma photon into ~38 visible photons per keV.
3. Light guide and PMT array - 37-91 photomultiplier tubes (increasingly position-sensitive PMTs or CZT solid-state detectors, which give direct conversion, better energy resolution and a compact camera).
4. Anger position logic - X and Y computed as the PMT-signal-weighted centroid; Z (energy) = sum of all PMT signals.
5. Pulse-height analyser (PHA) - a ~20% energy window around 140 keV rejects scattered (lower-energy) photons.
6. Corrections and display - energy, linearity and uniformity correction maps, then matrix (64², 128², 256²) framing.

SPECT acquisition and reconstruction

• One or more heads rotate around the patient collecting 60-128 planar projections over 360° (180° for cardiac, dual heads at 90°), step-and-shoot or continuous; 20-40 s per view.
• Projections stacked by slice form the sinogram; each row is the Radon transform of that slice.
• Reconstruction: filtered back projection (FBP) with a ramp filter rolled off by a Butterworth/Hann window, or iterative MLEM/OSEM - now standard, as it handles Poisson noise, attenuation, scatter and depth-dependent resolution (resolution recovery).
• Corrections: attenuation (Chang's uniform method, or a CT transmission map in SPECT/CT), scatter (dual/triple energy window), centre-of-rotation and motion.
• Gated SPECT MPI: ECG-gated into 8 or 16 frames per cardiac cycle to give ejection fraction and wall motion with perfusion.
• Typical clinical resolution: 8-12 mm FWHM; slice thickness 4-6 mm.

PET: coincidence detection and reconstruction

• Electronic collimation: two opposed detectors that register photons within a coincidence window τ of 2-6 ns define a line of response (LOR) - no lead collimator, so sensitivity is 2-3 orders higher than SPECT.
• Detector: a block of small scintillator crystals (2-4 mm pitch) read by PMTs or SiPMs (digital PET). Scintillators compared below; LSO/LYSO is today's standard - dense, fast, bright.
• Event types:
  • True coincidence - both photons from one annihilation, unscattered: the wanted signal.
  • Scattered coincidence - one or both photons Compton-scattered, so the LOR is wrong: adds background, corrected by model-based (single-scatter simulation) methods.
  • Random (accidental) coincidence - two unrelated photons within τ: rate R = 2τS₁S₂, grows as the square of activity; corrected by a delayed coincidence window.
  • Multiple events - three or more in the window, discarded.
• Time of flight (TOF) PET: if the arrival-time difference Δt is measured well (200-400 ps), the annihilation point along the LOR is localised to Δx = cΔt/2, improving effective SNR, especially in large patients.
• 2D vs 3D: septa-free 3D acquisition collects oblique LORs - much higher sensitivity, more scatter; reconstructed by 3D-OSEM, often after Fourier rebinning (FORE) to 2D.
• Corrections (all needed for quantification): normalisation (detector efficiency), dead time, randoms, scatter, attenuation from the CT (CT HU scaled to 511 keV μ), decay correction to injection time, and point-spread-function (PSF) modelling.
• PET/CT and PET/MR: CT gives the attenuation map plus anatomical localisation in one session; PET/MR offers better soft-tissue contrast but MR-based attenuation correction is harder. Long-axial-field-of-view ("total-body") PET raises sensitivity greatly.
• Typical clinical resolution: 4-6 mm FWHM, approaching 2-3 mm on digital systems.

Quantification, image quality and QC

• SUV (standardised uptake value) is the workhorse semi-quantitative index in FDG PET: tissue activity concentration normalised to injected dose per body weight (or lean body mass, SUL). Reported as SUVmax, SUVpeak, SUVmean; used for staging and treatment response (PERCIST/Deauville criteria).
• Factors that change SUV: uptake time after injection (standardise ~60 min), blood glucose and insulin, patient weight, reconstruction settings, lesion size (partial volume effect lowers SUV for lesions smaller than ~2-3 × FWHM), residual activity in the syringe, extravasation.
• Noise-equivalent count rate (NEC) is the single best figure of merit for PET count-rate performance; it peaks and then falls as activity rises because randoms grow quadratically.
• Performance standards: NEMA NU 1 for gamma cameras/SPECT, NEMA NU 2 for PET (spatial resolution, sensitivity, scatter fraction, NEC, image quality with the NEMA IEC body phantom), NEMA NU 4 for small-animal systems. IAEA and AAPM protocols are widely followed in Indian QA practice.
• Routine QC: gamma camera - daily energy peaking and uniformity (intrinsic and extrinsic), weekly spatial resolution/linearity (bar phantom), centre of rotation (COR) for SPECT, Jaszczak phantom for tomographic resolution/contrast/uniformity. PET - daily blank/normalisation check, detector gain, well counter calibration (cross-calibration of scanner with the dose calibrator, essential for SUV accuracy). Dose calibrator - constancy, accuracy, linearity, geometry.
• Artefacts: patient motion and PET/CT misregistration (breathing - a liver dome "banana" artefact), metal implants and contrast causing CT-based attenuation overcorrection, truncation at the CT FOV edge, brown fat and muscle FDG uptake, bladder and bowel activity, COR error producing ring/doughnut artefacts in SPECT.

Dose, safety and the Indian regulatory frame

• Internal dosimetry (MIRD outline): absorbed dose depends on cumulated activity à (area under the activity-time curve in the source organ) and an S-factor for each source-target pair; the governing time constant is the effective half-life, which combines physical decay and biological clearance.
• Typical effective doses (approximate, ICRP coefficients; verify against current ICRP publications): F-18 FDG ≈ 1.9 × 10⁻² mSv/MBq, so a 370 MBq study ≈ 7 mSv, plus 2-10 mSv for the CT; Tc-99m MDP ≈ 5.7 × 10⁻³ mSv/MBq, so 740 MBq ≈ 4 mSv.
• Dose reduction: correct tracer and activity for body weight, hydration and voiding, iterative reconstruction and TOF to allow lower activity, low-dose CT for attenuation correction only.
• Regulator: the Atomic Energy Regulatory Board (AERB), under the Atomic Energy Act, 1962 and the Atomic Energy (Radiation Protection) Rules, 2004. Licensing, registration and personnel approvals are done online through eLORA.
• Facility requirements (as set out in AERB's safety code for nuclear medicine facilities and its related guides - classified by the radionuclides and activities handled): layout approval, shielded hot lab with fume hood and lead bricks, L-block syringe shields, dose calibrator, contamination monitor and survey meter, delay-decay storage of radioactive waste, controlled/supervised areas, personnel monitoring by TLD/OSL badges, and an AERB-approved Radiological Safety Officer (RSO) of the appropriate level (RSO qualification via BARC's radiological physics / Dip.R.P. route). Equipment needs AERB type approval; transport of radioisotopes follows AERB/IAEA transport requirements. Specific code numbers and revisions change - check the current AERB document before quoting one.
• Patient release and advice after therapy doses (I-131, Lu-177) follows AERB dose-constraint criteria; diagnostic patients are advised on brief distancing from infants and pregnant contacts. Pregnancy is a relative contraindication; breastfeeding interruption is tracer-specific.

Key points

• SPECT - single gamma photons, mechanical lead collimator, Tc-99m 140 keV, resolution 8-12 mm, sensitivity ~10⁻⁴.
• PET - two 511 keV photons in coincidence, electronic collimation, resolution 4-6 mm, sensitivity 100-1000× SPECT.
• 511 keV comes from electron rest-mass energy; the two photons are ~180° apart, which is what makes coincidence possible.
• Tc-99m dominates SPECT: 6.01 h half-life, pure gamma at 140.5 keV, available daily from a Mo-99/Tc-99m generator.
• F-18 FDG dominates PET: 109.7 min half-life, glucose analogue trapped in metabolically active cells.
• Collimator resolution worsens linearly with distance from the collimator - keep the detector as close to the patient as possible.
• Resolution and sensitivity trade off in every collimator choice (LEHR vs LEAP vs pinhole).
• Anger logic - position = PMT-weighted centroid; energy = sum of PMT signals; the PHA window rejects scatter.
• Randoms rate = 2τS₁S₂ - quadratic in activity, so there is an optimum injected dose; corrected with a delayed window.
• NEC is the proper PET count-rate figure of merit, not raw prompt counts.
• TOF localises the event along the LOR: Δx = cΔt/2, about 6 cm for 400 ps - a noise, not a resolution, benefit.
• Attenuation at 511 keV is severe (factor of ~7 over 20 cm of water) and is corrected from the CT; without it PET is not quantitative.
• SUV = lesion activity concentration ÷ (injected activity/body weight); dimensionless, protocol-sensitive.
• Partial volume effect underestimates uptake in lesions smaller than about 2-3 × FWHM.
• Effective half-life 1/T_eff = 1/T_phys + 1/T_biol - always shorter than either.
• LYSO/LSO are today's PET crystals (fast, dense, bright); NaI(Tl) remains the SPECT crystal; CZT gives direct conversion.
• Theranostics pairs a diagnostic nuclide (Ga-68) with a therapeutic one (Lu-177) on the same ligand.
• AERB licenses every Indian nuclear medicine facility (eLORA) and requires an approved RSO, personnel monitoring and delay-decay waste handling.

Formulas and examples

Example 1 - F-18 decay between dispensing and scan. 555 MBq of FDG is dispensed at 07:00; the patient is scanned at 09:00. T½ = 109.7 min.

• t = 120 min → t/T½ = 120/109.7 = 1.0939
• A = 555 × 2^−1.0939 = 555 × e^{−1.0939 × 0.6931} = 555 × e^−0.7582 = 555 × 0.4685
• A ≈ 260 MBq - more than half gone, which is why F-18 doses are dispensed just before injection.

Example 2 - effective half-life of Tc-99m DTPA. Physical T½ = 6.01 h, biological clearance T½ = 24 h (renal).

• T_eff = (6.01 × 24)/(6.01 + 24) = 144.24/30.01
• T_eff ≈ 4.81 h - always less than the smaller of the two.

Example 3 - SPECT system resolution. LEHR collimator: d = 1.5 mm, L_eff = 40 mm; organ at b = 100 mm from the collimator face; intrinsic R_i = 3.8 mm.

• R_c = 1.5 × (40 + 100)/40 = 1.5 × 3.5 = 5.25 mm
• R_sys = √(5.25² + 3.8²) = √(27.56 + 14.44) = √42.0
• R_sys ≈ 6.5 mm FWHM. If the detector is moved to b = 200 mm, R_c = 1.5 × 240/40 = 9.0 mm and R_sys = √(81 + 14.44) = 9.8 mm - resolution degrades badly with distance.

Example 4 - attenuation and randoms in PET. (a) A 20 cm water path at 511 keV, μ = 0.096 cm⁻¹: transmission = e^{−0.096 × 20} = e^−1.92 = 0.1466, so ACF ≈ 6.8 - only ~15% of true coincidences survive. (b) Singles rates S₁ = S₂ = 1.0 × 10⁶ cps, τ = 6 ns: R = 2 × 6 × 10⁻⁹ × (10⁶)² = 2 × 6 × 10⁻⁹ × 10¹² = 12,000 cps. Halving τ to 3 ns halves randoms to 6,000 cps - hence the drive for fast crystals. (c) TOF Δt = 400 ps: Δx = 3 × 10⁸ × 400 × 10⁻¹² / 2 = 0.12/2 = 0.06 m = 6 cm FWHM along the LOR.

Example 5 - SUV of a lung lesion. Injected 370 MBq FDG (decay-corrected to scan time), body weight 70 kg, measured lesion concentration 40 kBq/mL.

• A_inj/W = 370,000 kBq / 70,000 g = 5.286 kBq/g ≈ 5.286 kBq/mL (tissue density ≈ 1 g/mL)
• SUV = 40 / 5.286 = 7.57 - well above the ~2.5 figure often used as a rough malignancy threshold, though the threshold is protocol-dependent and not diagnostic on its own.
• If the same lesion were only 8 mm across in a scanner with 5 mm FWHM, the partial volume effect could depress the measured SUV by 30-50%.

Where it is used

• Oncology - FDG PET/CT for staging, restaging and response assessment of lymphoma, lung, head-and-neck, oesophageal and breast cancer; Ga-68 PSMA for prostate cancer and Ga-68 DOTATATE for neuroendocrine tumours, routine at Tata Memorial, AIIMS, RGCI, HCG and most large private networks.
• Cardiology - Tc-99m sestamibi/tetrofosmin gated SPECT myocardial perfusion for ischaemia and ejection fraction; FDG PET for myocardial viability; N-13 ammonia/Rb-82 for absolute myocardial blood flow.
• Bone, renal, thyroid and lung studies - Tc-99m MDP bone scans for metastases, MAG3/DTPA renograms for obstruction and split function, pertechnetate/I-123 thyroid scans, V/Q scans for pulmonary embolism; high-volume routine SPECT work in government medical colleges.
• Neurology and psychiatry - FDG and amyloid/tau PET for dementia, ictal/interictal SPECT for epilepsy focus localisation, DaTSCAN-type dopamine transporter imaging for Parkinsonism.
• Industry and job roles - nuclear medicine technologist and medical physicist/RSO in hospitals; radiopharmacy and cyclotron operations at BRIT, Jubilant Radiopharma and hospital cyclotron units; field service and applications engineers for GE Healthcare, Siemens Healthineers and Philips PET/CT and SPECT/CT systems; reconstruction and AI image-analysis roles in imaging software firms; regulatory and QA roles interfacing with AERB eLORA submissions.
• Research - small-animal PET/SPECT (NEMA NU 4) in pharmacology and radiopharmaceutical development at BARC, IITs and national institutes; dosimetry for Lu-177 and I-131 therapy planning.

Interview questions

• Q: Why does PET not need a collimator? A: Because the two annihilation photons travel in nearly opposite directions, detecting them in coincidence defines the line of response electronically - this is why PET sensitivity is two to three orders of magnitude higher than SPECT.
• Q: Why is the PET photon energy exactly 511 keV? A: It is the rest-mass energy of the electron (and of the positron); the annihilation converts both rest masses into two photons of 511 keV each.
• Q: What makes Tc-99m almost ideal for SPECT? A: A 6 h half-life matching clinical workflow, a single 140.5 keV gamma with no particulate emission (low patient dose), efficient detection in a thin NaI crystal, and daily availability from a Mo-99/Tc-99m generator with versatile labelling chemistry.
• Q: How does collimator choice affect the image? A: Smaller, longer holes improve spatial resolution but cut sensitivity, so counts fall and noise rises for the same time - every collimator is a resolution-versus-sensitivity compromise; hole size must also match the photon energy to limit septal penetration.
• Q: What limits PET spatial resolution fundamentally? A: Positron range before annihilation, non-collinearity of the two photons (±0.25°, worse with larger ring diameter), detector crystal width and block decoding error, plus reconstruction smoothing.
• Q: What are random coincidences and how are they corrected? A: Two photons from different annihilations arriving within the coincidence window; the rate is 2τS₁S₂ (quadratic in activity) and is estimated with a delayed coincidence window and subtracted, or estimated from singles rates.
• Q: Why is CT part of a PET/CT scanner? A: It provides the attenuation map (CT HU scaled to 511 keV) for quantitative attenuation correction and gives anatomical localisation of uptake in the same session - a low-dose CT suffices if diagnostic CT is not needed.
• Q: What is SUV and why should you not trust a single value? A: SUV is lesion activity concentration divided by injected activity per body weight; it depends on uptake time, blood glucose, weight normalisation, lesion size (partial volume), reconstruction settings and scanner cross-calibration, so it is only comparable within a standardised protocol.
• Q: What does time-of-flight PET improve - resolution or noise? A: Primarily noise: knowing Δt localises the event along the LOR (Δx = cΔt/2, ~6 cm at 400 ps), which reduces the propagation of noise in reconstruction, most noticeably in heavy patients. Spatial resolution is largely unchanged.
• Q: A gated cardiac SPECT shows a ring-shaped artefact in every slice. First suspect? A: A centre-of-rotation error or a non-uniformity in the detector head - check the COR calibration and the extrinsic uniformity flood.
• Q: What is effective half-life and why does it matter more than physical half-life? A: It combines physical decay with biological excretion (1/T_eff = 1/T_phys + 1/T_biol); it governs cumulated activity and hence the absorbed dose to the patient and the precautions after discharge.
• Q: What are the main regulatory requirements to run a PET/CT facility in India? A: AERB licence through eLORA under the Atomic Energy Act, 1962 and the Atomic Energy (Radiation Protection) Rules, 2004, with layout approval, type-approved equipment, an AERB-approved RSO, personnel monitoring badges, a shielded hot lab, delay-decay radioactive waste handling and documented QA - the applicable AERB safety code and its current revision should be checked before quoting it.

Common mistakes

• Saying PET uses a collimator, or that SPECT uses coincidence detection - the collimation mechanism is the single defining difference.
• Confusing the radionuclide with the radiopharmaceutical: F-18 is the nuclide, FDG is the labelled molecule that decides where it goes.
• Using physical half-life for dose calculations when biological clearance also applies - the effective half-life is required.
• Forgetting to decay-correct the injected activity to the scan time before computing SUV, or ignoring residual activity in the syringe.
• Mixing units - mCi with MBq (1 mCi = 37 MBq), kBq/mL with Bq/g, ns with ps in randoms and TOF formulas.
• Quoting collimator resolution as a fixed number: it worsens linearly with source-to-collimator distance, so detector proximity matters.
• Reading any high FDG uptake as cancer - inflammation, infection, brown fat, muscle and physiological brain, myocardium, bowel and bladder activity all light up.
• Treating a negative scan as proof of absence - small lesions below the partial-volume limit and non-FDG-avid tumours exist.

Related topics

X-ray imaging basics · CT and MRI basics · Ultrasound basics · Medical image processing · Radiation safety and AERB · Biomedical instrumentation and radiation detectors