2020/04/07 by Stefan Denner, Denner, Stefan, Ashkan Khakzar +11
Medicine · #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Image and Video Processing (eess.IV) #Multiple Sclerosis Research Studies #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2004.03675
openalex publication_date 2020/04/07 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Segmentation of Multiple Sclerosis (MS) lesions in longitudinal brain MR\nscans is performed for monitoring the progression of MS lesions. We hypothesize\nthat the spatio-temporal cues in longitudinal data can aid the segmentation\nalgorithm. Therefore, we propose a multi-task learning approach by defining an\nauxiliary self-supervised task of deformable registration between two\ntime-points to guide the neural network toward learning from spatio-temporal\nchanges. We show the efficacy of our method on a clinical dataset comprised of\n70 patients with one follow-up study for each patient. Our results show that\nspatio-temporal information in longitudinal data is a beneficial cue for\nimproving segmentation. We improve the result of current state-of-the-art by\n2.6% in terms of overall score (p<0.05). Code is publicly available.\n