Abstract: Recent advancements in computer vision and image processing have significantly transformed healthcare, enhancing diagnostic precision, reducing costs, and improving efficiency. Among medical imaging techniques, Magnetic Resonance Imaging (MRI) stands out for its capability to identify even minute brain abnormalities. This study presents a comparative analysis of two advanced object detection models, YOLOv5 and YOLOv7, for brain tumor detection and classification using MRI scans. The dataset includes three major tumor categories—meningioma, glioma, and pituitary tumors. Preprocessing techniques and mask alignment methods were applied to enhance segmentation accuracy before model training.
Experimental evaluation shows YOLOv5 achieved a recall of 0.905 for box detection and 0.906 for mask segmentation, with a precision of 0.94 and 0.936 respectively. At an IoU threshold of 0.5, it attained a mean Average Precision (mAP) of 0.947, while YOLOv7 achieved slightly higher accuracies with 0.936 and 0.935 in detection and segmentation. YOLOv7 also produced better mAP scores across varying IoU ranges. Comparative analysis with traditional models such as RCNN, Faster RCNN, and Mask RCNN further confirms the efficiency and reliability of YOLO-based architectures for accurate brain tumor identification.
Keywords: Brain Tumor, Deep Learning, Image Processing, MRI, YOLO, Object Detection, Segmentation, mAP, RCNN.
Downloads:
|
DOI:
10.17148/IJARCCE.2025.141127
[1] Anitha L, Harshitha B S, Apoorva B M, Manasa G B, Annie Shreya D, "Automated Brain Tumor Segmentation and Classification in MRI Using Yolo-Based Deep Learning," International Journal of Advanced Research in Computer and Communication Engineering (IJARCCE), DOI: 10.17148/IJARCCE.2025.141127
