A Novel Approach for Deep Learning Based Brain Tumor Classification Using MATLAB and MRI Imaging

early and accurate bram tumor diagnostics. The interpretation of MRI scans by hand is usually quite time-consuming and can be facilitated by various errors, so there is a need to create automated diagnostic devices. A deeper learning-based brain tumor classification based on a customized Convolutional Neural Network (CNN) in MATLAB is described in this paper. The publicly available MRI datasets trained the model, and the forms of tumors were glioma, meningioma, and pituitary tumors. With an accuracy of 92, the proposed model was more efficient than some of the other earlier models, including ResBRNet (89) and EfficientNet V2 (87), which indicates the efficacy of the proposed architecture. The effectiveness of the model was also enhanced based on data augmentation approach like rotation, flipping, and shifting that played a vital role in reducing overfitting and enhancing generalization. The model proved to be quite accurate, recalls and F1-scores in tumor classification, and however, there are still significant difficulties in the differentiation between glioma and pituitary tumours. Another strength of the model noted that was in its capacity to generate high values of AUC, thus confirming its strength.