A Novel Approach to Brain Tumor Detection Using Texture Based Gabor Filter Followed by Genetic Algorithm

Abstract

It is very important to consider the factors like blurred boundaries and external noise while analyzing the brain tumor structures. It is difficult to segment accurately the brain MRI. Many studies in both developing and developed countries indicate that an inactive diagnosis has led to the death of the majority of people who suffer from brain tumor. The proposed novel method is a filter mechanism based on texture Gabor filter, which is followed by a genetic algorithm proposed to improve segmentation accuracy. A texture-based Gabor filter has been used to detect irregularities and to extract statistical properties further used in segmentation and classification. In order to improve segmentation effectiveness, a better separation of different clusters of the features from Gabor filter is studied. An objective function has been also formalized to adjust filter parameters with gradient descent and genetic algorithm. This document has demonstrated the effects of the segmentation of both qualitative and quantitative productivity. The findings show that the novel proposed approach works better with respect to the segmentation accuracy.