Photons reaching two opposed detectors flow along a timeline; if a photon falls in the other detector within the window around a photon, a coincidence is recorded: true if from the same annihilation, random if from different ones. Rates by window width and activity are plotted.
Key point: true coincidences rise linearly with activity and level off once the window covers the arrival difference; randoms rise linearly with window width and with the square of activity, so noise grows at high activity and with a wide window.
Source: book (T ∝ A, R = 2τ·S₁·S₂ ∝ A², window 6–12 ns conventional, ~3–6 ns TOF; scatter 30–40% in 3D). The window width is shown as 2τ (Cherry et al.). Model: photons of one annihilation arrive at most ±2 ns apart (~60 cm patient), timing uncertainty σ 0.25 ns; singles Poisson. On the timeline the time between events is shortened for display (in reality ~10⁵–10⁶ singles/s per detector); curves relative.
How to use: a single click on the drawing pauses and resumes; ↑ ↓ change the window width in the first step and the activity in the second; ← → change steps.
Coincidence timing window and count rates · Open in Nuclear Medicine Explained ↗