A Comparative Study of Deep Learning Algorithms in Classifying Brain Cancer

Cancer remains a significant global health concern, contributing to a substantial number of fatalities. Specifically, brain cancer (BC) entails the abnormal proliferation of cells within or near the brain. Timely detection plays a pivotal role in effective tumor treatment. Machine learning advancements have significantly aided the medical field in diagnosing brain tumors. Convolutional Neural Network (CNN) has emerged as a widely utilized and effective machine learning technique for image processing and medical imaging. Nonetheless, many existing classification algorithms exhibit inconsistencies. In recent years, deep learning has exhibited promising outcomes in BC classification. This study focuses on the usage of four popular deep learning algorithms, namely AlexNet, GoogLeNet, InceptionNetV3, and ResNet50, utilizing brain Magnetic Resonance Imaging (MRI) slices to classify cancerous tumors into Glioma Tumor (GT), Meningioma Tumor (MT), Pituitary Tumor (PT), and No Tumor (NT) categories. The performance of these algorithms has been thoroughly assessed using metrics such as accuracy, precision, recall, and F1 score. Notably, GoogLeNet demonstrates superior results among the four models employed. Additionally, a comparative analysis against various state-of-the-art classification algorithms highlights the enhanced performance of these deep learning models.