Certain investigations on the performance analysis of mri brain tumor classification using machine learning and deep learning models

Abstract

Cancer is a significant cause of death globally and is expected to newlinesurpass other causes in the coming years. Approximately 40,000 to 50,000 newlinecases of brain tumors are diagnosed annually in India, with approximately newline20% being minors. The International Association of Cancer Registries newline(IARC), 2022 reports that over 24,000 people die yearly from brain tumors in newlineIndia. The report says that the brain cancer market is expected to increase at a newlinecompound annual growth rate of 1.11 % till 2030 and brain tumor might newlinebecome the second-most common cancer by 2030 in India. Early detection newlineand treatment of brain cancer are the most effective ways to decrease newlinemortality rates. Despite years of research, brain tumors are still one of the newlinemost lethal forms of cancer, resulting from abnormal cell growth in the brain newlineor its supporting tissues that can damage or compromise brain function. With newlineadvancing technology, Computer-Aided Diagnostic (CAD) tools are used to newlineidentify brain tumors predict prognosisand assess reoccurrence likelihood. In newlinedigital histopathology, an essential aspect of CAD systems involves analyzing newlinecellular images, drawing from the research in computer vision. newlineThis thesis focuses on developing three innovative Computer-Aided newlineDesign (CAD) techniques that employ machine learning and deep learning newlinemethods to detect and categorize brain tumors in MRI images into newlineMeningioma, Gliomaand Pituitary tumors. The primary objective of this newlineresearch is to create, executeand assess a pattern recognition system that can newlineimprove the precision of brain tumor classification. newline