Comparative evaluation of EPI and SWI for the assessment of PRL and CVS in Multiple Sclerosis

Abstract Introduction The 2024 McDonald criteria incorporate the central vein sign (CVS) and paramagnetic rim lesions (PRL) as supportive imaging biomarkers for MS diagnosis. While susceptibility-weighted-imaging (SWI) and T2*-weighted echo-planar-imaging (EPI) are generally used to assess CVS/PRL, their relative performance remains unclear. This study compared high-resolution isotropic-T2*-EPI with non-isotropic SWI for CVS/PRL detection. Materials and Methods In this multi-centre study, 21 patients with MS underwent harmonized 3T-MRI including EPI and SWI. CVS and PRL were evaluated according to NAIMS criteria. Whole-brain and controlled lesion analyses on 120 pre-selected lesions were performed independently for each contrast, with EPI serving as reference standard. Results In whole-brain analyses, SWI showed good sensitivity for CVS eligibility and positivity (AC1=0.68-0.78) but significant directional disagreement with EPI (p<0.0001). Discrepancies were primarily attributed to limited lesion-parenchyma contrast and venous visibility on SWI, which improved using low-flip-angle SWI. Controlled lesion analyses supported these observations. For PRL, SWI demonstrated high sensitivity (88%) and precision (97%) compared to EPI, though systematic bias persisted (p<0.001). Controlled lesion analyses showed more balanced, albeit moderate performance. Conclusion SWI diverged systematically from EPI for CVS and PRL detection. When available, EPI should be preferred, while optimised low-flip-angle SWI may serve as an alternative to conventional SWI.