Multiple Sclerosis Lesion Detection and Classification by Using Support Vector Machine SVM

Multiple sclerosis (MS) is an autoimmune malady of the central nervous system. To interpret this disease and its progression, magnetic resonance imaging (MRI) is increasingly exploited. This paper focuses on the segmentation and the detection of MS lesions by cerebral magnetic resonance imaging. Lately, machine learning (ML) techniques have been used in biomedicine and computer science to make diagnostic and analytical decisions to fight multiple sclerosis. We propose the SVM approach to analyze MRI images for the medical application of Multiple Sclerosis. The subject of this study is to assess this approach on a real public database, the proposed technique has been tested in the BraTS 2020 dataset which contains 369 denoting MRI images of patients associated with MS. Each one includes 4 MRI images modalities which are T1 weighting (T1), T2 weighting (T2), Fluid Attenuated Inversion Recovery (FLAIR) and T1 with gadolinium-enhancing contrast (T1EC). This paper is concerned with verifying the effectiveness of this algorithm by comparing these results against the reference performed by specialists, we further compared our segmentations with those of FSL tools, Freesurfer and statistical Parametric Mapping (SPM) since these methods are considered to be the best tools for brain segmentation. The propound approach is validated by using Python program.