Feature Extraction from Brain MR Images for Detecting Brain Tumor using Deep Learning Techniques

Abstract

Detection of a brain tumor due to their intricacy, the irregularity of their tumor formations, and the variety of their tissue textures and forms, gliomas provide a difficult problem for medical image interpretation. Machine learning-based approaches to semantic segmentation have consistently surpassed earlier techniques in this difficult challenge. However, some of the Machine learning techniques are unable to deliver the necessary local information associated with changes in tissue texture brought on by tumor development. In this study, we used a Hybrid technique that combines supervised learning features and handcrafted features. The texture features based on the grey level co-occurrence matrix (GLCM) are used to build the hand-crafted features. The recommended technique also lowers the intensity of nearby unimportant areas and only the region of interest (ROI) method is used, which precisely represents the input size of the entire tumor structure. ROI MRI scan pixels are divided into several tumor components using a decision tree (DT).