YRT-PET-NX: Development of a Plugin for an Open-Source PET Reconstruction Platform to Support the NeuroEXPLORER

Le résumé fourni par la source

Ultra-high resolution PET scanners offer multiple benefits to brain imaging for researchers and clinicians. The NeuroEXPLORER (NX) scanner, a brain-dedicated PET system with high resolution and high sensitivity, achieves exceptional image quality. These benefits, however, introduce challenges for image reconstruction in terms of speed, memory usage and need for accurate system modeling and motion correction. Open-source third-party image reconstruction tools can be a valuable substitute to manufacturer software to address these challenges, as well as to enable the evaluation of different modeling approaches and algorithms. In this work, the opensource PET reconstruction platform YRT-PET (Yale Reconstruction Toolkit for Positron Emission Tomography) is extended through a plugin to support the NX scanner. YRT-PET provides a scanner-agnostic reconstruction platform with research-oriented features. The YRT-PET-NX plugin integrates detector geometry, listmode packet decoding and scanner-specific corrections. We evaluated the performance by reconstructing a mini-Derenzo phantom, a non-human primate, and a human brain, comparing YRT-PET results with those from the vendor software and the MOLAR engine. Based on the mini-Derenzo reconstructions, comparable resolution was achieved, with a slight gain in peak-to-valley ratios. Across both non-human primate and human datasets, YRT-PET produced nearly-identical images. Time-activity curves for the non-human primate were quantitatively consistant, with a 5 - to 18 -fold reduction in reconstruction time due to GPU acceleration. These results demonstrate that the YRT-PET-NX plugin enables efficient and highquality reconstructions comparable to preexisting and established software. This work enables the features of YRT-PET alongside the high resolution of the NX. Future work aims to expand support for additional corrections such as randoms and scatter estimation.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.