Breast masses detection on mammograms using recent one-shot deep object detectors
Ella Mahoro, Moulay A. Akhloufi · 2023
Breast cancer is a prevalent cancer that affects mostly women worldwide. The detection of breast masses is a crucial part of breast cancer diagnosis and treatment. However, breast masses detection on mammograms is challenging due to variations in breast tissue appearance and abnormalities’ complexity. This article presents a breast mass detection system that utilizes recent deep object detection techniques such as YOLOv7 and YOLOv8. The system’s first step involves extracting images from the VinDr-Mammo dataset and enhancing image quality using different techniques like contrast-limited adaptive histogram equalization, median filter, and bilateral filter. With a mean average precision of 0.65, YOLOv8 outperformed YOLOv7, demonstrating the system’s potential to detect masses on mammograms.