A design of a brain tumor classifier of magnetic resonance imaging images using ResNet101V2 with hyperparameter tuning

Abstract

Brain tumors are a disease that is quite dangerous and requires severe treatment. One thing that is quite important is the process of diagnosing the brain tumor. This diagnosis process requires intense attention, and differences in interpretation may arise. Machine learning has been used in several fields, including disease diagnosis. This paper proposes an intelligent diagnostic tool for brain tumors using ResNet101v2. ResNet101V2 is used to classify meningioma, glioma, pituitary, and normal from magnetic resonance imaging (MRI) images. This research includes data collection, data preprocessing, ResNet101v2 design and evaluation. We investigate three models of ResNet101v2 for brain tumor classification. The best model achieves an accuracy of 96.2%.