Answers are grounded in this paper and cite their sources.
Inter-crystal scatter often distributes the energy of an annihilation photon across two crystals and leaves two candidate lines of response (LORs) that conventional detectors cannot reliably distinguish. Dividing the count between the two lines avoids this decision, but existing proportional weights depend on the activity distribution of each acquisition. Here we derive the weight from the detector geometry instead. In GATE simulations of the PET coil brain insert, we tabulated from a 180 s normalization acquisition the probability that the crystal with the larger energy is the first interaction crystal, against crystal separation and opposing crystal position. The probability fell from 0.40 for adjacent crystals to 0.05 at 14.6 mm and differed by a factor of 1.8 between the two sides of the scattering crystal at oblique incidence. Two implementations were evaluated: a geometry-dependent table and its event-weighted mean as a constant weight. Both implementations improved LOR count accuracy over conventional assignments in uniform, hot–cold and Hoffman phantoms and performed equally across them. After 1200 s of the uniform phantom, the table gave a count error of 22.4%, against 26.9% for the lower energy assignment. In Hoffman images, it raised the gray matter signal-to-noise ratio from 5.09 to 7.52.
No takes yet. Share an insight, caveat, or question.
Shim et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: