Enhanced Brain Tumor Detection via ResNet50 Classification and Residual U-Net Segmentation

AI-driven dual-model system for brain tumor detection in MRI scans, combining classification and segmentation techniques. A ResNet50-based convolutional neural network (CNN) identifies the presence of tumors, while a ResUNet architecture ensures precise segmentation for accurate localization. To enhance localization, ResUNet architecture is employed for pixel-wise segmentation, providing precise tumour boundaries within the MRI images. By incorporating transfer learning techniques, the models benefit from pre-trained weights, significantly improving performance and reducing training time. Innovative use of Focal Tversky loss addresses data imbalance, improving segmentation accuracy for irregular tumour shapes. Experimental results demonstrate superior performance with 95% classification accuracy and a Dice score of 0.91 for segmentation. This dual-model pipeline offers a robust solution for early detection and precise localization of brain tumours, which is critical for timely medical intervention and improved patient outcomes.