An interpretable CNN framework for multi-class brain tumor classification with visual explanation in MRI scans

Brain tumors are among the most life-threatening neurological conditions, and effective therapy and patient outcomes depend on accurate and timely diagnosis. This study proposes an interpretable and deployment-oriented deep learning framework for multi-class classification of brain tumors using magnetic resonance imaging (MRI). The framework employs a carefully designed lightweight CNN configuration that prioritizes computational efficiency and robustness for practical clinical use. Evaluations on a publicly available MRI dataset demonstrated the effectiveness of the proposed framework, achieving a classification accuracy of 99.54%. To enhance clinical trust, Grad-CAM visualizations were integrated to highlight class-discriminative regions within MRI scans, providing transparent and interpretable predictions. In addition, the framework was deployed within a user-friendly desktop application, enabling real-time diagnostic support. Comparative analysis with state-of-the-art models shows that the proposed system achieves competitive or superior performance while maintaining efficiency, making it a promising candidate for clinical decision support, subject to further validation in real-world settings.