Cutting-Edge Deep Learning Models for Enhanced Breast Can-cer Detection: Comparative Analysis of YOLOv5, YOLOv8 and YOLOv9 Models

Abstract

Breast cancer is a leading cause of mortality among women worldwide, necessitating advancements in early and accurate detection methods. This study explores the efficacy of cutting-edge deep learning techniques, specifically YOLOv5, YOLOv8, and YOLOv9, for breast cancer detection. The YOLO (You Only Look Once) family of algorithms, known for their speed and precision, were employed to identify cancerous lesions in mammographic images. In our comparative analysis, YOLOv5 achieved a precision confidence curve of 1.00 at 0.786, an F1 confidence curve of 0.76 at 0.377, a precision-recall curve with an mAP@0.5 of 0.770, and a recall confidence curve of 0.92 at 0.000. YOLOv8 demonstrated improved performance with a precision confidence curve of 1.00 at 0.583, an F1 confidence curve of 0.97 at 0.575, a precision-recall curve with an mAP@0.5 of 0.969, and a recall confidence curve of 0.96 at 0.000. YOLOv9 showed the highest effectiveness with a precision confidence curve of 1.00 at 0.657, an F1 confidence curve of 0.97 at 0.467, a precision-recall curve with an mAP@0.5 of 0.974, and a recall confidence curve of 0.97 at 0.000. These results underscore the potential of advanced deep learning models in enhancing breast cancer detection. Future work will focus on refining these models to improve their robustness and applicability across diverse clinical settings. The integration of these techniques into routine screening processes could significantly advance early detection and treatment outcomes for breast cancer patients.