Brain Tumor MRI Classification using Xception Model

The proper classification from magnetic resonance imaging (MRI) scans of a brain tumor is essential for a correct and timely diagnosis, thus improving above all patients’ outcomes. This research proposes an advanced classification framework utilizing the Xception model, integrated with transfer learning, to classify brain tumors into four categories: Meningioma, No Tumor, Glioma, and Pituitary. To perform successful model training, robust data preprocessing techniques such as resizing, normalization and augmentation are employed. The model achieved high accuracy of 91.01% and precision in the training, validation and test datasets, generalization and a good reliability. Analysis through a confusion matrix gives class wise performance, and takes ability to interpret system from probabilities as validation. The model was deployed as a web-based application (Streamlit) object producing real time predictions and clinical usable web-based application. Results show the promise of AI enabled solutions in reducing diagnostic accuracy and operational efficiency in medical imaging.