Heightened Breast Cancer Segmentation in Mammogram Images

Breast cancer stands as the foremost form of malignant tumor identified in women, and the successful treatment hinges on its timely detection. The assessment derived from mammogram images continues to serve as the benchmark for diagnosing breast cancer. The intricate characteristics of breast lesions depicted in mammogram images pose challenges in reliably accomplishing segmentation. This paper introduces a novel automated quantitative image analysis technique for mammogram images. Indeed, in recent decades, convolutional neural networks (CNNs) have risen as a promising and influential methodology, demonstrating the capability to attain state-of-the-art results across various applications in medical imaging fields. To enhance the recovery of comprehensive localization information, several segmentation techniques based on mammogram images have recently been integrated with CNN methods. This study presents a computerized approach based on an Enhanced Seg-Net architecture for the segmentation of breast masses, aiming to contribute to the diagnosis of breast cancer. The findings indicate that the Enhanced U-Net outperforms alternative testing methods, achieving a precision of 97.12%, a recall of 95.03%, and an F1-score of 98.06%, all accomplished within a reduced inference time. In comparison to state-of-the-art techniques, the proposed method produces a higher accuracy rate by specifying breast cancer location.