Enhanced Evaluation of Juxtacortical Lesions in Multiple Sclerosis Using High-Resolution Double Inversion Recovery Imaging with Deep Learning

Motivation: Accurate detection of juxtacortical lesions in MS within a clinically feasible timeframe remains challenging due to the long scan times required by 3D double inversion recovery (DIR) imaging. Goal(s): This study aimed to evaluate the efficacy of high-resolution DIR imaging enabled by the new deep learning-based technique, DL-Speed (DLS-DIR). Approach: The study conducted a retrospective analysis of MRI data from 25 MS patients using C-DIR and DLS-DIR. Two neuroradiologists independently assessed juxtacortical lesions, comparing lesion counts across methods and calculating interobserver agreement. Results: DLS-DIR significantly outperformed C-DIR, especially in detecting cortical MS lesions, within a clinically acceptable acquisition time. Impact: 3D DL-Speed double inversion recovery imaging is acquired in 4:23 with 0.7mm isotropic resolution covering the whole brain, which could improve the accuracy of cortical lesion assessment in patients with multiple sclerosis.