Automated brain MRI image classification through machine learning and deep learning approaches

Abstract Tumors within the human brain cause severe effects on the cerebral system. Therefore, the early detection of such abnormalities needs to be accurately performed, and classifications of such cells also play a pivotal role in planning further treatment procedures. Timely detection and classification of brain tumors remain challenging, motivating the investigation of machine learning and deep learning approaches for computer-assisted image analysis. To serve the purpose, Support Vector Machine, Logistic Regression, Random Forest, U-Net, and k-Nearest Neighbors models for the identification and correct categorization of the brain tumors have been considered. The models have been trained and tested using the brain MRI images. The considered models are implemented in Python, and the outcomes of the models are observed critically to select the best model for the computer-aided diagnosis system. Among the models, Random Forest and k-Nearest Neighbors have illustrated the highest classification accuracy and computational efficiency of 94.44% and 93.89%, with runtimes of 0.073s and 10.97s, respectively, in correct identification and categorization of the brain tumor through MRI images.