Brain Tumor Segmentation using proposed YOLO algorithm

Improving survival rates for brain cancers requires early discovery and precise diagnosis, with MRI scans being critical to identifying anomalies. Traditional tumor segmentation methods are often time-consuming and require manual adjustments, which hinder real-time diagnosis. This study introduces the proposed YOLOv11 model, an advanced deep learning approach designed to enhance brain tumor segmentation. Proposed YOLOv11 integrates CSPDarkNet53 in its backbone for detailed feature extraction and utilizes Path Aggregation Networks (PAN) in the neck architecture to merge multi-scale features, ensuring accurate segmentation of both small and large tumors. Unlike conventional methods, P- YOLOv11 directly generates segmentation masks from learned features, eliminating the need for post-processing steps and improving efficiency. Experimental results demonstrate that YOLOv11 achieves a mAP of 99.2% for object detection and tumor segmentation, outperforming previous YOLO versions such as YOLOv8 and YOLOv10. Key performance metrics, including precision, recall, and confidence curves, validate its high accuracy.