DSU-Net: An Attention-Enhanced Dense Skip U-Net for Breast Lesion Segmentation in Mammographic Images
Researchers have developed DSU-Net, a novel deep learning model designed to improve the segmentation of breast lesions in mammographic images. This attention-enhanced Dense Skip U-Net architecture aims to assist radiologists by providing more accurate and consistent lesion localization, which is crucial for early breast cancer detection. The model demonstrated strong performance on the CBIS-DDSM dataset, achieving a Dice Similarity Coefficient of 0.9421 and an AUC-ROC of 0.9878, indicating its potential to enhance computer-aided screening and diagnosis. AI
IMPACT Improves accuracy in medical image analysis, potentially leading to earlier and more reliable breast cancer detection.