Detecting Brain Tumour from MRI images using multifeature analysis convolution neural network

Abstract

A brain tumor is a serious health problem; around 170000 patients are affected by this disease. According to World Health Organization (WHO) report, primary brain tumors affect around 2% of people. The brain tumor creates several health difficulties, such as speech problems, affecting thinking ability, confusion, and behavior changes. Therefore, the brain tumor must be recognized earlier to eliminate the mortality rate. Various researchers use radiographic images to analyze brain tumors using machine learning techniques. The traditional methods consume high computation time while exploring the high dimensional feature set. In addition, the brain image noise removal process consumes high computation time, leading to overfitting issues. Hence, this study uses the Multi-Feature Analysis Convolution Neural Model (MFCNM) to identify the brain tumor at an earlier stage. Initially, the brain MRI images are processed using the Histogram equalization approach that enhances the image quality and can handle the non-patch and patch regions. Then correlated features are extracted to segment the tumor region, which is done by using the Fuzzy assimilated clustering approach. The fuzzy-based segmented region solving the uncertainty issues and classification process is performed with the help of Feature Map based Transform Model that identifies the tumor region with minimum analyzing and computation time. Then the effectiveness of the system is evaluated using experimental results. newline