MRI brain tumor segmentation and prediction using modified region growing and adaptive SVM

Abstract

Magnetic resonance imaging (MRI) is one of the tumor diagnostic tools in any part of the body. Nowadays, the brain tumor is becoming a major cause of the death of many individuals. The seriousness of a brain tumor is very big among all the variety of cancers, so to save a life immediate detection and proper treatment to be done. Detection of these cells is a difficult problem, because of the formation of the tumor cells. It is very essential to compare brain tumor from the MRI treatment. It is very difficult to have a vision of the abnormal structures of the human brain using simple imaging techniques. To overcome a problem, in this paper, automated brain tumor detection and classification approaches are proposed. The proposed work consists of five stages, namely preprocessing, segmentation, feature extraction, feature selection, and classification. In the first step, preprocessing is performed to extract the region of interest (ROI) using manual skull stripping, and noise effects are removed by the median filter. Then, the tumor is segmented in the second step by improved modified region growing algorithm (MRG), which contains both orientation constrains and intensity constraints. Then, GLCM-based texture features are extracted in the third step. After that, the best features are selected by the grasshopper optimization algorithm (GOA). Finally, the adaptive support vector machine (ASVM) is to classify types of tumors. Experimental results are analyzed in terms of different metrics. Results and experiments show that the proposed method accurately segments and classifies the brain tumor in MR images.