Automated segmentation of multiple sclerosis lesion in intensity enhanced flair MRI using texture features and support vector machine

Abstract

In this paper, a fully automated segmentation method is proposed to identify Multiple Sclerosis (MS) related white matter lesions from brain magnetic resonance imaging (MRI) data. The main contribution of this paper is to obtain a new texture feature set for MS Lesion segmentation that is a combination of local and global neighbourhood information. The proposed method adopts a robust intensity normalization technique and lesion contrast enhancementfilter for enhancing the region of interest. We use a Support Vector Machine (SVM) to classify lesion pixels and level set based active contour and morphological filtering to achieve higher accuracy on lesion pixel identification. Quantitative evaluation of the proposed method is carried on real MRI data set provided by MS Lesion Challenge 2008. The results obtained from our method indicate significant improvement in performance compare to three state of the art methods that shows the proposed method's high suitability for assisting the neurologist to detect the MS in clinical practice.