A Hybrid Framework-Integrating a Radiomatic Features with CNN for Brain Tumor Detection

Brain tumors are life-threatening diseases, and early diagnosis using MRI is crucial for effective treatment. This study presents an interpretable framework for brain tumor detection using MRI images. A lightweight Convolutional Neural Network (CNN) is used to classify images as tumor or non-tumor, while a traditional Gray Level Co-occurrence Matrix (GLCM) with Support Vector Machine (SVM) serves as a baseline model. The CNN achieves higher accuracy than the GLCM+SVM approach. To improve transparency, Grad-CAM is applied to generate heatmaps that highlight the regions influencing the CNN's predictions. The results demonstrate that combining deep learning with explainable AI can enhance both diagnostic accuracy and model interpretability in MRI-based brain tumor detection.