BrainSeg3D to Detect Multiple Sclerosis Lesions Using Magnetic Resonance Imaging

摘要

Today, many researches and many existing methods treat the multiple sclerosis (MS) which is a chronic inflammatory pathology of the central nervous system. To detect multiple sclerosis outliers, we should have a diagnosis by magnetic resonance imaging (MRI) with different sequences. For this purpose, some recent methods of segmentation and classification of MRI sequences have been proposed for the automatic detection of MS tumors. In this paper, we propose a new protocol based on the existing and novel semi-automated lesions segmentation methods of BrainSeg3D. This protocol is applied on a novel MR databases of 30 MS patients, which were acquired with a 3T MR scanner with conventional sequences in order to evaluate our test with the obtained Consensus segmentation. The obtained results show an improvement comparant to those proposed in the literature.