Implementation of Brain Tumor Detection and Classification Model Using CNN Approach

Unless diagnosed and treated early, brain tumors unusual growths may prove to be lethal. Even with the standard methods, such as MRI scans, to precisely diagnose brain cancers, it may be expensive, time-consuming, and reliant on specialized interpretation. In this research, CNNs trained on MRI image datasets that are employed to suggest an autonomous brain tumor classification framework. The proposed model categorizes four distinct brain states as meningioma, glioma, pituitary tumor, and no tumor utilizing the approach transfer learning together with the Xception architecture. For generalization of the model, the dataset was pre-processed utilizing image augmentation. We evaluated the performance of model using metrics such as F1-score, recall, accuracy, and precision. Following refinement, the final model achieved 99.7% efficiency on the test dataset. The suggested method has promise for use in clinical settings by drastically cutting down on the time and skill needed for brain tumor diagnosis.