Abstract
386
Objectives: Development of brain dedicated PET scanners, and other organ specific or unconventional geometries, requires high performance detectors capable of providing high XY resolution, timing resolution, DOI information and high sensitivity simultaneously. It is well known that one-to-one coupling between scintillator pixels and photodetector elements is favorable over light sharing schemes in terms of timing performance, however such approach lacks DOI information unless dual end readout is used and also results in XY resolution limited to the photodetector size. One approach to achieve DOI information is to stack multiple scintillator crystal arrays with an offset to separate pixels from different depths in the flood map. We are exploring the possibility of using this staggered approach, but have one-to-one coupling between the bottom crystal layer and the photodetector array, in order to achieve a combination of DOI information and timing resolution.
Methods: As a first step a series of light transport simulations have been performed using the Monte Carlo code DETECT2000 to compare characteristics of one-to-one coupling, traditional light sharing, and offset crystal stacks (2 and 3 layers with equal thickness) with and without light sharing (4 crystal pixels per MPPC pixel). In all cases the total detector thickness was 20 mm LYSO:Ce. 8x8 Multi Pixel Photon Counter (MPPC) arrays were simulated with pixel size of 2 and 3 mm. Rise time and decay constants were τr=68 ps, τd1=21.5 ns (13%) and τd2=43.8 ns (87%). Gamma-ray events (@511 keV) with 15000 optical photons per event were simulated in the center of the detector throughout the crystal depth. The photodetection efficiency of the MPPC array was 45% and the absorption length in the crystal 40 cm. The light distribution on the MPPC array was studied as a function of interaction depth, and the FWHM of the distribution of time of arrival of the 5th and 10th detected optical photon was calculated as a metric for timing performance. The timing performance was studied using both the maximum MPPC signal and the summed signals across the array for the time stamp.
Results: The light spread over the MPPC array showed that for our proposed novel staggered geometries it is straightforward to separate events from each crystal layer based on the light distribution on the MPPC array. For 2 layers of 2 mm pixels coupled to an MPPC array with the same pixel size, for events in the bottom layer 68% of the detected light will go to to the main MPPC pixel and in the top layer 86% of the light will be shared between two adjacent pixels. Corresponding values for a 3 layer configuration is 62%, 2x33% and 4x21%. In terms of timing performance one-to-one coupling was found superior to light sharing. For the one-to-one staggered detector timing resolution could be preserved in the layer closest to the MPPC array when the maximum signal was used for triggering. We observed 180 and 184 ps FWHM for 2 and 3 layers, compared to 150 ps for one layer with one-to-one coupling. Corresponding values for staggered with light sharing was 250 and 270 ps. However, timing resolution degraded with distance from the MPPC array. For the summed signal the opposite behavior was observed, and timing improved with distance from MPPC in the layered detectors. For the summed signal differences were more subtle across the detector configurations compared to the maximum signal.
Conclusions: The current results show promise in that several levels of DOI in combination with good timing resolution at least in part of the detector can be achieved. Current work includes experimental verification of the described approach. For this study we are using PETsys electronics for readout and investigate performance as a function of number of layers as well as layer thicknesses, which will be presented at the SNMMI conference. Acknowledgment: This work was funded by NIH grant R21EB023391.