Timing Calibration with Depth of Interaction for the NeuroEXPLORER Brain PET

The NeuroEXPLORER (NX) is a next generation SiPM based dedicated brain PET scanner. It offers ultra-high sensitivity, spatial resolution, and an extended axial field of view. To provide previously unachievable imaging performance, NX uses novel U-shaped LYSO crystals with reduced sizes to provide depth of interaction (DOI) and time of flight (TOF) measurements. Timing calibration is critical to achieve the best possible timing resolution. In this work, we investigate timing calibration with DOI for the NX. We develop an iterative algorithm to accurately estimate time offsets on a per crystal and DOI basis using a line source. We first estimate the source position using, e.g., a non-TOF reconstruction. The difference time error for each event is computed by comparing the measured time and true time obtained based on the source position, which in turn is used to generate error histograms for each DOI of each crystal. The mean is estimated from a Gaussian fitting, which is used as a candidate for updating time offsets using an iterative algorithm. To further reduce data acquisition time, we introduce a separable time offset model (STOM), the time offsets are mainly contributed by crystal and DOI independently. The STOM allows timing calibration incorporating DOI information while reducing unknowns comparable to those in a non-DOI timing calibration. The overall mean timing resolution in FWHM is 271.3±13.9ps. The difference between STOM time offsets and full estimated time offsets is quite small; the standard error is about 1 TOF bin of 12.21ps. In summary, we develop effective timing calibration with DOI using a line source for the NX.