Tumour semantic segmentation and SVM classification using magnetic resonance imaging data

Abstract

A brain tumour is a malignant growth of abnormal cells or tissues that causes damage to the brain. Brain tumour tissue detection is extremely challenging when examining the entire brain. The key to successful treatment is early tumour discovery. In modern times, detecting and segmenting the brain-tumour area from brain MRI pictures using detection or segmentation methods has proven to be a very helpful strategy. Magnetic resonance imaging is a challenging area of image processing because of the critical importance of precision in medical diagnostics. The supplied material includes one MRI scans. Brain tumour segmentation is the process of extracting MRI pictures of the brain from normal brain tissue. Pre-processing steps for MRI scans include applying filters like median filtering and cranium removal before submitting them to a thresholding procedure using the watershed segmentation technique. The tumour area is then divided and acquired. Then, in later phases, GLCM techniques implemented in MATLAB were used to retrieve the characteristics of interest. Support vector machine (SVM) was then used to classify the pictures; the algorithm achieved an average accuracy of 93.05%. Which is significantly superior to the usual versions.