Detection and diagnosis of glioma brain tumor using machine learning and modified visual geometry group CNN model

Abstract

The Glial cells in human brain region affected by abnormal development and newlinegrowth of the cells and leads to Glioma formation. Glioma tumors are mostly occurred in newlinecerebrum of the human brain and 80% of all tumors formed in human brain are Glioma newlineand the survival rate of the Glioma tumor case is about 12 to 18 months. Headache and newlinesevere seizure are the common symptoms of Glioma tumors. Continuous vomiting and newlinevision loss are the later symptoms of Glioma tumors. All age of people are affected by newlinethese Glioma tumors but mostly men are highly affected than the women. The Glioma newlinetumors can be categorized into Low Grade Glioma (LGG) and High Grade Glioma (HGG) newlinebased on the location of the tumor tissues and their size. The LGG can be formed in the newlinebrain region due to the following two cells as astrocytes and oligodendrocytes newlineThese limitations are overcome by proposing deep learning algorithms for the newlineGlioma detection process. The deep learning structures LeNET and AlexNET methods newlineare applied on the source brain images to detect the Glioma brain image category. The newlinedeep leraning architecture in general methodology consists of Convolutional Layer newline(CLayer) and Down Sampling Layer (DS_Layer) and Fully Connected Neural Networks newline(FCNN) with different set of internal neurons in each layers. The LeNET structure used in newlinethis design consist of CLayer1 and CLayer2 with DS_Layer1 and DS_Layer2 and three newlineFCNN layers as FCNN1, FCNN2 and FCNN3 respectively. newlineThe Glioma detection rate and tumor region segmentation accuracy was not newlineoptimum in the above methods for further tumor diagnosis process. Therefore, there is newlinea need for the system which performs both Glioma image detection and Glioma tumor newlinediagnosis with high tumor region segmentation accuracy. Hence, the Modified Visual newlineGeometry Group (MVGG) architecture is proposed to detect and diagnose the tumors in newlineGlioma images newline