QSM-RimDS: A highly sensitive paramagnetic rim lesion detection and segmentation tool for multiple sclerosis lesions

Motivation: Paramagnetic rim lesions (PRLs) are imaging biomarker of the innate immune response in MS lesions. QSM-RimNet can identify PRLs but requires precise QSM lesion mask and does not provide rim segmentation. Goal(s): To develop QSM-RimDS algorithm to detect PRLs using the readily available FLAIR lesion mask and to provide rim segmentation for microglial quantification. Approach: We trained an nnUNET for PRL rim segmentation and classified PRL status based on rim length relative to FLAIR lesion perimeter. Results: QSM-RimDS achieved 90% detection sensitivity at 53.3% precision vs. 24.7% by QSM-RimNet. Compared to expert segmentation, 77% of PRLs have a DICE score greater than 0.5. Impact: QSM-RimDS has the potential to replace human readers for the time-consuming PRL detection and segmentation task while improving reproducibility.