An optimized DOI decoding method for a PET detector with nine-crystal-to-one-photodetector coupling.

Positron Emission Tomography (PET) benefits from depth-of-interaction (DOI) information to correct parallax errors introduced by thick scintillator crystals, especially in dedicated brain systems where edge-of-field DOI effects are most pronounced. Conventional light-sharing window (LSW) readout is limited to a four-to-one or two-to-one crystal-to-photodetector coupling and cannot decode DOI for central crystals under nine-to-one coupling schemes. This work proposes a novel detector architecture that overcomes these limitations, enabling DOI decoding for all crystals and supporting coupling ratios up to sixteen-to-one. Approach: We designed a LYSO scintillator array (1×1×20 mm³ crystals in a 12×15 matrix) coupled to a 4×5 Hamamatsu MPPC array, incorporating transparent-interface LSWs. A nine-to-one coupling configuration was implemented. Using collimation experiments and interpolation, a transfer function linking DOI to the MPPC signal ratio (ROM) was constructed. The detector performance was evaluated using flood maps and energy spectra, and the DOI mean absolute error (MAE) was calculated at interaction depths of 4 mm, 8 mm, 12 mm, and 16 mm in collimation experiments. Main results: The detector achieved an energy resolution of 11.8%. DOI MAEs at 4, 8, 12, and 16 mm were 2.34, 2.71, 3.73, and 5.37 mm, respectively, yielding an average MAE of 3.54 mm. DOI decoding was successfully performed for all crystals, including those centrally coupled. Compared to conventional LSW-based configurations, the detector achieved more than a 1.5-fold improvement in spatial resolution. Significance: The transparent-interface LSW design enables consistent DOI decoding across all scintillator elements at elevated coupling ratios and can be extended to support coupling schemes of up to sixteen-to-one. Moreover, by reducing pixel dimensions and enabling full-array DOI extraction, this architecture delivers superior positioning accuracy and spatial resolution, making it ideally suited for small animal PET and dedicated brain PET applications.