Image Segmentation on MRI images for brain Tumor identification using AI and Deep Learning approach

Abstract

On contrast enhanced T1 weighted magnetic resonance (MR) images, a cellular automaton (CA) based seeded tumour segmentation method is proposed, which standardises the volume of interest (VOI) and seed selection. The CA algorithm is examined in this paper to establish a link between CA-based segmentation and graph-theoretic methods, demonstrating that the iterative CA framework solves the shortest path problem with proper transition rule selection. Tumor-cut segmentation to partition the tumour tissue further into necrotic and enhancing parts, and later probability maps constructed from each modality could be combined to obtain the final segmentation and smoothing by level set active surfaces. After that, SVM classification is performed using the final segmented image feature extraction and feature selection process. The Classified result will distinguish between normal and abnormal brain images and will be robust in terms of computation time, tumour type heterogeneity, and efficiency.