Development of Robust Framework for Automatic Segmentation of Brain MRI Images

Abstract

AbstractThe robust framework for automatic segmentation of brain MRI images is developed in this work. The brain MRI image has been typically used to analyzing the brain anatomical structures and disorder in the brain. The identification of viable segmentation framework for brain MRI image is notoriously difficult problem. The major difficulty in brain MRI image analysis is mostly due to the unattainable of efficient segmentation framework. The proposed work develops a robust framework for classification and segmentation of brain MRI images. The developed robust framework consists of two phases, i.e., classification of input MRI images and detection of tumor. The automatic SVM classification and improved watershed methods are used in the robust framework to improve the segmentation process and improve the segmentation accuracy. In the first phase, MRI images classification is obtained with training and testing process with SVM classifier. In the second phase, segmentation of tumor region in MRI image is presented with improved watershed method. The developed robust framework is quantitatively evaluated on benchmark brain MRI image dataset.KeywordsBrain tumorSVM methodWatershed methodRobust frameworkBilateral filterBrain MRI image