The effect of TOF versus non-TOF on defective PET detectors in clinical simultaneous ¹⁸F-FDG PET/MR Imaging

Objectives Spontaneous detector failure may occur unpredictably, even when routine quality checks are normal. Continuation of scanning may be necessary in case a detector failure occurs, e.g. when patients are already injected with radiotracer, or when the exam cannot be postponed. In literature, only sparse information can be found about the effect of defective detectors on PET image quality. Moreover, the research published in literature was performed on conventional PET/CT scanners with photomultiplier tubes, while PET/MR scanners (and some recent PET/CT scanners) have different detector types (solid-state photodetector devices) and layouts. Two major solid-state photodetectors groups are: the avalanche photodiodes (APDs), and the silicon photomultipliers (SiPMs). The latter is ‘faster’ and could also provide time-of-flight (TOF) information. Together, this makes it difficult to decide whether or not the patients can be reliably scanned on PET/MR systems with a defective detector. Therefore, the objective of this study was to describe and evaluate the influence of TOF vs. non-TOF reconstructions on clinical PET image quality in simultaneous PET/MR imaging with defective PET detectors.

Methods A total of 6 patients with malignant tumors were included and underwent 18F-FDG TOF PET/MRI. TOF and non-TOF PET images were reconstructed for reference and after simulating 1, 3, and 5 defective detector units in 4 different locations in the detector ring. The whole-body scans were divided in three regions: head and neck, chest, and abdomen. In all three regions the images were visually assessed and scored in three categories: artifacts, overall image quality, and reader confidence. A note was made when the artifacts would lead to a change in diagnosis. In addition, a quantitative assessment was performed. Percentage error maps and cumulative error distribution functions were calculated. Differences were ascertained and compared using the Wilcoxon matched pairs signed-ranks test.

Results Without the inclusion of TOF information the image artifacts introduced by one defective detector unit degraded the overall image quality significantly (p=0.03), reduced the confidence (p=0.03) and could lead to a change in diagnosis (25% probability). With the inclusion of TOF information artifacts were reduced (p=0.03). Images reconstructed with one defective detector unit had similar scores as the images reconstructed without defective units. Three or five defective detector units introduced more artifacts, reduced the overall image quality and confidence (p=0.03), but less severe compared to non-TOF images. The absolute percentage error for one, three and five defective detector units was 8%, 20%, and 37%, respectively, for the non-TOF cases, and 5%, 11%, 19%, respectively, for the TOF cases (p=0.03). Without TOF there was a probability of 58% that the absolute error in a voxel would be less than 10%. With TOF this probability was increased to 72%.

Conclusions Our study indicates that PET image artifacts elicited by defective detectors are significantly mitigated with the integration of TOF information in simultaneous PET/MR. One defective detector unit yields an absolute percentage error of approximately 5%. However, in TOF-imaging, even in cases with one defective unit, overall image quality, artefacts and reader confidence are not significantly degraded.