U-Net-Driven Advancements in Breast Cancer Detection and Segmentation

Breast cancer remains one of the most prevalent malignancies affecting women worldwide, necessitating advancements in diagnostic methodologies to improve patient outcomes. Traditional imaging techniques, while effective, often require enhanced computational tools to assist in accurate diagnosis and treatment planning. In this study, we proposed a novel neural network architecture specifically designed for breast cancer segmentation and classification in ultrasound images. The model features dual-phase encoding and decoding stages, augmented with skip connections and dropout regularization, to optimize feature extraction and reduce overfitting. Evaluation on a comprehensive dataset demonstrates exceptional performance metrics, with an accuracy of 0.9887 and a dice coefficient of 0.9227, affirming the model’s robustness and efficacy in detecting and classifying breast cancer tumors. Collaboration with medical professionals for extensive validation will enhance the model’s interpretability and clinical applicability, ultimately improving early diagnosis and treatment outcomes for breast cancer patients.