Fast Maximum Likelihood Positioning for a Staggered Layer Scintillation PET Detector
Christoph W. Lerche, Wenwei Bi, Mirjam Schoeneck, Debora Niekaemper, Qi Liu, Elisabeth Pfaehler, Lutz Tellmann, Juergen J. Scheins, N. Jon Shah
TL;DR
This work introduces an iteration-free Maximum Likelihood Positioning (MLP) algorithm tailored for staggered-layer, pixelated scintillation detectors in PET, enabling simultaneous estimation of gamma energy and identification of the active scintillator pixel with depth-of-interaction information. By approximating the log-likelihood with a maximum-normalized, per-pixel distribution and implementing SIMD-accelerated, multi-threaded processing, the method achieves up to $\sim 22.5\times 10^{6}$ singles/s on a 60-thread platform, while maintaining energy resolution around $12\%$ FWHM after energy correction. The approach is validated on a 3-layer BrainPET 7T detector (1634 pixels) with calibration from $^{68}\mathrm{Ge}$ sources, achieving accurate pixel identification and energy reconstruction despite light-yield variations and boundary effects. While most events are correctly assigned, some edge-case ambiguities arise from the staggered geometry, highlighting both the practicality and limitations of DOI-enabled, high-throughput MLP in this detector design. Overall, the work demonstrates a fast, scalable MLP framework suitable for high-rate PET imaging with DOI-capable staggered-layer detectors.
Abstract
In this study, we propose a fast implementation of a Maximum Likelihood Positioning (MLP) algorithm to estimate the energy and identify the active scintillator pixel in staggered layer scintillation detectors for PET. The staggered layer design with pixelated scintillators enables the determination of the gamma's depth of interaction and facilitates an iteration-free formulation of the MLP algorithm. The efficacy of the algorithm optimization was tested on a scintillation detector block designed for an ultra-high field BrainPET 7T, comprising three scintillator pixel layers. The three layers contain 24 x 24, 24 x 23 and 23 x 22 scintillator pixels, respectively, with a pixel pitch of 2 mm in both directions and layer thicknesses of 9, 8 and 7 mm. Calibration measurements, in combination with an automated calibration script, were used to obtain the expected counts of scintillation photons required in the MLP algorithm. Using Single-Instruction-Multiple-Data parallelization, multi-threading and optimized cache lines, a maximum processing speed of approximately 22.5 million singles per second was achieved on a platform with four Intel Xeon Platinum 8168 CPUs and 60 threads, encompassing all required processing steps. The automatic calibration failed for 1 to 15 individual scintillator pixels in approximately 10 per cent of the 120 scintillation detector blocks, necessitating manual correction. After applying the energy correction to the positioned single events, an energy resolution of of 12 +/- 2 per cent FWHM was obtained for the entire scintillation block. This value is very close to the energy resolutions measured for the individual scintillator pixels, proving that the MLP accurately identifies the scintillating pixel and that the energy correction method effectively compensates for the light collection variations of the SiPM array.