Integrating YOLOV8 and Grad-CAM++ for Enhanced Brain Tumor Detection and Interpretation

Early diagnosis of brain tumors is essential for improved patient outcomes. However, traditional MRI-based diagnostics heavily rely on manual radiological analysis, which is often difficulties in latency and inaccuracies. This work used the integration of yolov8 and Grad-cam++ to enhance the accuracy and interpretation of tumor identification from an image dataset. YOLOV8 demonstrated 91% accuracy in real-time tumor localization, thereby confirming its speed and precision for clinical application. Grad-CAM++ has improved transparency of models with a heat-map with 99% accuracy, showing significant areas graphically relevant for authentication by medical professionals about the forecast. The technology is quite efficient with the application of these techniques along with CNN architectures like ResNet50 model. This new approach addresses both technological and clinical constraints, simultaneously offering a scalable, interpretable, and efficient diagnostic tool for identification of brain tumor.