Certain investigations on glioma Brain tumor detection and diagnosis Using eml nhmm and deep learning Architecture

Abstract

The detection of tumor regions in brain images are done by either newlineinvasive or non-invasive method. In case of invasive method, the foreign newlinematerial is inserted into the human brain which locates the abnormal regions newlinein brain. This method consumes more time for the tumor region detection and newlinealso produces high pain for the patients. The blood loss is inevitable in this newlinemethod. These limitations are tolerated by proposing non-invasive method for newlinedetecting and locating the tumor regions in brain. This non-invasive method is newlinebased on the scanning techniques, which can be categorized into Computer newlineTomography (CT) and Magnetic Resonance Imaging (MRI). In this thesis, newlineMRI scanning technique is used to detect and segment the tumor regions. newlineIn this research work, the brain tumors are detected and diagnosed newlineusing machine learning approaches. The noise variations in brain images are newlinedetected and removed using index filter, which is proposed in this research newlinework. The noise filtered images are transformed into multi orientation based newlinebrain image using Gabor transform method. Then, the hybrid features which newlineare the integration of statistical and texture features, are computed from this newlinetransformed brain image. These computed features are classified using EML newlineapproach, which categorize the source brain image into either normal or newlineabnormal image. Then, the segmented tumor regions are diagnosed using newlineCANFIS classifier, which classifies the segmented regions into mild or newlinesevere. newline