Sub‐2 mm depth of interaction localization in PET detectors with prismatoid light guide arrays and single‐ended readout using convolutional neural networks

Abstract

Purpose Depth of interaction (DOI) readout in PET imaging has been researched in efforts to mitigate parallax error, which would enable the development of small diameter, high‐resolution PET scanners. However, DOI PET has not yet been commercialized due to the lack of practical, cost‐effective, and data efficient DOI readout methods. The rationale for this study was to develop a supervised machine learning algorithm for DOI estimation in PET that can be trained and deployed on unique sets of crystals. Methods Depth collimated flood data was experimentally acquired using a Na‐22 source with a depth‐encoding single‐ended readout Prism‐PET module consisting of lutetium yttrium orthosilicate (LYSO) crystals coupled 4‐to‐1 to 3×3 mm2 silicon photomultiplier (SiPM) pixels on one end and a segmented prismatoid light guide array on the other end. A convolutional neural network (CNN) was trained to perform DOI estimation on data from center, edge and corner crystals in the Prism‐PET module using (a) all non‐zero readout pixels and (b) only the 4 highest readout signals per event. CNN testing was performed on data from crystals not included in CNN training. Results An average DOI resolution of 1.84 mm full width at half maximum (FWHM) across all crystals was achieved when using all readout signals per event with the CNN compared to 3.04 mm FWHM DOI resolution using classical estimation. When using only the 4 highest signals per event, an average DOI resolution of 1.92 mm FWHM was achieved, representing only a 4% dropoff in CNN performance compared to using all non‐zero pixels per event. Conclusions Our CNN‐based DOI estimation algorithm provides the best reported DOI resolution in a single‐ended readout module and can be readily deployed on crystals not used for model training.

Depth of Interaction and Coincidence Time Resolution Characterization of Ultrahigh Resolution Time-of-Flight Prism-PET Modules

“…The best timing resolution performance of Prism-PET modules in head-on configuration with DOI filter (4-to-1 Prism-PET module, ∼254 ps) is close to that of the state-of-the-art TOF PET scanner, Siemens Biograph Vision [30], although investigation and verification at the system level should be implemented in the near future. However, it is significant to note that the PET module with single-ended readout and uniform glass light guide has achieved up to 157 ps FWHM CTR using depth correction techniques [18], which we have looked into by exploring the potential of the machine-learning using a convolutional neural network (CNN) in training our collected data to improve the DOI resolution and hence achieve better CTR [19], [23], [31].…”

High‐resolution and high‐sensitivity PET for quantitative molecular imaging of the monoaminergic nuclei: A GATE simulation study

“…Both scanners consist of 14 detector rings and 40 detector blocks per ring, forming an axial FOV of 37.2 cm. Such a compact geometry is only possible thanks to the high DOI resolution of the Prism‐PET modules 65,69 …”

“…The ground truth DOI was calculated based on the distance from the hit position to the surface of the corresponding crystal. In order to accurately simulate the DOI performance of the Prism‐PET detector modules, Gaussian noise with 4 and 2 mm full width at half maximum (FWHM) was added to the ground truth DOI of the Prism‐PET 1 mm and 1.5 mm scanners, respectively 65,68,69 …”

“…In order to accurately simulate the DOI performance of the Prism-PET detector modules, Gaussian noise with 4 and 2 mm full width at half maximum (FWHM) was added to the ground truth DOI of the Prism-PET 1 mm and 1.5 mm scanners, respectively. 65,68,69 A virtual cylindrical scanner (VC) with 0.5 × 0.5 mm 2 virtual crystals was modeled, shown as the yellow ring in Figure 2. Each list-mode event was rebinned to a pair of VC crystals along the same line of response (LOR).…”

Development and Evaluation of a Dual-Layer-Offset PET Detector Constructed with Different Reflectors

“…(2) Arranging a light guide at the top of the crystal arrays to guide the optical photons to specific photodetectors and thus decode the DOI. LaBella et al designed and fabricated a novel prismatoid light-guide array that realized a DOI resolution of 2.5 mm and an energy resolution of 9% [29,30]. Continuous crystals are also used in small animal PET scanners [31].…”