Unmasking VGG16: LIME Visualizations for Brain Tumor Diagnosis

Brain tumors are complex and potentially life-threatening conditions requiring accurate diagnosis. This study explores using VGG16 CNN architecture to classify brain tumors in MRI scans. CNNs excel in image classification but lack transparency, raising concerns in critical applications like medical diagnosis. To address this, we use Explainable AI (XAI) techniques. We enhance interpretability using Superpixel images, Heatmaps, and Image Masks to visualize important features for predictions. Our approach achieves 98.8% accuracy, with visualizations aiding in understanding the model’s decisions and providing insights for medical professionals. This research contributes to XAI in medical imaging, showing the potential of combining deep learning with interpretable methods for healthcare decision-making.