TOF PET/CT Scanner Timing Calibration Using a Stationary Line Source and Lutetium Background

384 Introduction: Adding time-of-flight (TOF) technology has been proven to improve image quality in positron emission tomography (PET). In order for TOF information to significantly reduce the statistical noise in reconstructed PET images, good timing resolution is needed across the scanner field of view (FOV). An accurate, fast and convenient timing calibration method is necessary to achieve such a timing resolution. Existing timing calibration methods either require a large extended extrinsic radiation source or a rotating line source, or require long data acquisition time from intrinsic radiation. This study describes method for lutetium-based scanners using a stationary, centered line source together with lutetium background radiation. Timing offset and timing walk coefficients are calibrated for each individual crystal in order to optimize the timing performance for a TOF PET scanner. Methods: A prototype, SiPM-based TOF Canon PET/CT scanner was used for the timing calibration algorithm development in this study. The scanner consisted of 48 circularly-arranged detector units (DUs), each consisting of an array of 12 x 60 LYSO crystals. Timing offset and timing walk coefficients are calibration parameters which are used to convert the measured time stamp of an individual detected gamma ray to a calibrated time stamp, including an energy-dependent term. Our timing calibration method consists of two steps: the first using a centered line source and the second using lutetium background. In the first step, data were collected for 3 minutes from a 2.4-mCi, 68Ge line source at the center of the FOV. The number of coincidence events was estimated to be enough to calibrate the peak position from the TOF difference histogram for each crystal. From these data, two calibration parameters were determined: the relative timing offset coefficients within opposing DU pairs, and timing walk coefficients for each crystal. Relative timing offset coefficients were calibrated from per-crystal TOF difference histograms through an iterative procedure. Walk correction coefficients were then calibrated by nonlinear fitting to the TOF difference versus energy curve. In the second step, data were collected for 6 minutes from the detector’s intrinsic lutetium background. The number of coincidence events only needed to be enough to calibrate the peak position of the TOF difference histogram for each DU pair. Timing offset coefficients between DUs were calculated analytically from histograms of the TOF differences among each pair of DUs. Because timing walk coefficients had been previously applied, events with a wide range of energies from lutetium background data could be used, reducing the required number of total counts from intrinsic radiation. A combined timing offset coefficient was calculated for each crystal as the sum of the relative crystal timing offset coefficients within DU pair in step (1) and the timing offset coefficients between DUs in step (2). The combined coefficients were applied to coincidence data from a separate validation experiment. Results: After correcting relative timing offset and timing walk coefficients, peak positions from DU pair TOF difference histograms are clearly identified for lutetium events with a wide range of energies. The timing resolution averaged over all events in the validation dataset was 255 ps FWHM for 68Ge line source at scanner isocenter. Conclusions: In this work, an accurate, convenient and fast timing calibration method using a centered line source and lutetium background radiation was developed. The proposed method requires no large extended phantoms or motorized phantoms. Total data acquisition time is only several minutes with a 2.4 mCi 68Ge line source.