Deep Learning-Based Brain Tumor Detection Using MRI with Explainable AI
Abstract
Tumor detection from MRI images is a crucial step that needs precision and fast diagnosis in order to yield positive results for the patient. Tumor classification via MRI scans manually by radiologists takes a lot of time and is vulnerable to human mistakes. In order to solve such problems, this research project suggests an automatic system for brain tumor detection using deep learning techniques. A transfer learning technique based on Xception framework is used in order to categorize MRI images into one of four classes: pituitary tumor, glioma, meningioma, and no tumor. In order to provide interpretability of model predictions, Gradient-Weighted Class Activation Mapping (Grad-CAM) method is implemented for identification of the most significant part of the image that helped the neural network come up with the decision. This way, Grad-CAM provides tumor localization. The solution also involves an automatic reporting system which presents model predictions in a structured manner. The proposed model achieves an overall accuracy of 99%, with precision and recall values of 0.99 across all tumor classes, demonstrating its effectiveness as a diagnostic aid tool.