Brain tumor classification model using convolutional neural networks on magnetic resonance imaging

Abstract

Accurate classification of brain tumor images is an important challenge in the field of healthcare, demanding fast and reliable diagnoses for proper treatment. In this article, we propose a model that enhances for the automated classification of brain tumor images brain tumor classification, aiming to support diagnostics, increase confidence, and reduce time in the classification process of meningioma, glioma, and pituitary tumors, in magnetic resonance imaging (MRI) scans. Our results demonstrated that the VGG16 model achieved the best classification results among the three evaluated models. Its deep architecture and learning capacity allowed it to learn more discriminative representations of brain tumors, resulting in a higher accuracy rate. Furthermore, the results of the VGG16 implementation were compared to previous studies representing the state of the art in the field. This comparison highlighted the effectiveness and relevance of the VGG16 model for this classification problem. These results offer valuable insights for future development of automated brain tumor detection systems.