An automatic Multiple Sclerosis lesion segmentation approach based on cellular learning automata: A case study

Abstract

Multiple Sclerosis (MS) is a demyelinating nerve disease which for an unknown reason assumes that the immune system of the body is affected, and the immune cells begin to destroy the myelin sheath of nerve cells. In Pathology, the areas of the distributed demyelination are called lesions that are pathologic characteristics of the Multiple Sclerosis (MS) disease. In this research, the segmentation of the lesions from one another is studied by using gray scale features and the dimensions of the lesions. The brain Magnetic Resonance Imaging (MRI) images in three planes (T1, T2, PD) (weighted) containing MS disease lesions have been used. Cellular Learning Automata (CLA) is applied on the MRI images with a novel trial and error approach to set penalty and reward frames for each pixel. The images were analyzed in MATLAB and the results show the MS disease lesions in white and the brain anatomy in red on a black background. The proposed approach can be considered as a supplementary or superior method for other methods such as Graph Cuts (GC), fuzzy c-means, mean-shift, k-Nearest Neighbor (KNN), Support Vector Machines (SVM).