Abstract:
Because of the intricate nature of tumor formations and their neurological effects, brain tumor identification and segmentation continue to be significant issues in medical imaging. With an AlexNet-based CNN for classification and YOLOv8 for segmentation, this study offers a sophisticated deep learning architecture that achieves 96% classification accuracy and 93.8% mAP@0.5 for segmentation. Our system, which was trained on 5,064 annotated MRI scans from Kaggle, performs better than earlier iterations of YOLO and is especially strong at detecting meningiomas (98.5% mAP). The methodology provides a scalable solution for clinical procedures by drastically cutting down on analysis time without sacrificing diagnostic accuracy. This study highlights how automated, trustworthy tumor analysis using AI-driven diagnostics can improve healthcare outcomes.
