Physics, Instrumentation, and Radiation Detection
Physics and instrumentation
Nuclear imaging physics reduces to three interaction modes. Photoelectric absorption dominates at diagnostic gamma energies in high-Z detector material and produces the full-energy photopeak; its probability scales roughly with Z³/E³. Compton scatter dominates in soft tissue at 100–511 keV and is why energy discrimination and scatter windows exist. Pair production requires 1.022 MeV and matters mainly for high-energy therapy isotopes.
In NaI(Tl) the crystal yields about 38 photons per keV deposited, and the photomultiplier chain amplifies roughly 10⁶-fold so pulse height is proportional to deposited energy. Energy resolution of a clinical gamma camera is about 6–7% FWHM at the 140-keV photopeak, which is why a 20% window (±10%, 126–154 keV) is customary. Higher-energy photons need thicker septa; high-resolution collimators buy resolution by giving up sensitivity, and no reconstruction recovers counts that were never detected.
PET relies on coincidence detection of opposed 511-keV annihilation photons. Modern block detectors (LSO/LYSO, BGO, GSO) differ in light output, decay time, and stopping power; time-of-flight improves signal-to-noise in large patients. Randoms grow as the product of singles rates, and dead time always makes the reported count rate underestimate the true rate at high activity.
Common traps: believing a larger reconstruction matrix improves resolution; choosing a high-sensitivity collimator for a small lesion; forgetting that CZT semiconductor cameras offer better energy resolution but different sensitivity and geometry; and confusing randoms with scatter.
Worked example: for a nonparalyzable system, τ = 1/m − 1/n. Observed 80,000 cps against a true 100,000 cps gives τ = 1/80,000 − 1/100,000 = 2.5 × 10⁻⁶ s, i.e. 2.5 µs - a 20% count loss.