SCAR-Net-assisted ultrasound diagnosis of postoperative scars and recurrent lesions in breast cancer.
Authors
Affiliations (8)
Affiliations (8)
- Department of Diagnostic Ultrasound Imaging & Interventional Therapy, Zhejiang Cancer Hospital, Hangzhou Institute of Medicine (HIM), Chinese Academy of Sciences, Hangzhou, China.
- Department of Ultrasound, Shaoxing People's Hospital (Zhejiang University Shaoxing Hospital), Shaoxing, China.
- The First Affiliated Hospital of Zhejiang Chinese Medical University (Zhejiang Provincial Hospital of Chinese Medicine), Hangzhou, China.
- Interventional Medicine and Engineering Research Center, Hangzhou Institute of Medicine (HIM), Chinese Academy of Sciences, Hangzhou, China.
- Zhejiang Provincial Research Center for Innovative Technology and Equipment in Interventional Oncology, Zhejiang Cancer Hospital, Hangzhou, China.
- The First People's Hospital of Lin'an District, Hangzhou, China.
- Wenzhou Medical University, Wenzhou, China.
- Department of Medical Engineering, Zhejiang Cancer Hospital, Hangzhou Institute of Medicine (HIM), Chinese Academy of Sciences, Hangzhou, China.
Abstract
Postoperative differentiation between scar tissue and recurrent lesions in patients with breast cancer presents a significant diagnostic challenge. This study introduces SCAR-Net, a deep learning model specifically designed for ultrasound-based discrimination between these similar-appearing tissues. Using 34,376 ultrasound images from 5,710 patients across four hospitals, we developed a model incorporating scar-recurrence feature enhancer and boundary-sensitive attention network modules. In multicenter validation, SCAR-Net significantly improved radiologists' diagnostic performance, increasing AUC from 0.810-0.824 to 0.939-0.942, sensitivity from 0.775-0.782 to 0.934-0.941, and specificity from 0.839-0.872 to 0.935-0.950 (all <i>p</i> < 0.001). The model also demonstrated superior segmentation accuracy with Dice coefficients of 0.918-0.925 compared to traditional methods (0.795-0.811). These results suggest SCAR-Net could serve as a valuable auxiliary tool for improving early recurrence detection in postoperative breast cancer follow-up, potentially reducing unnecessary biopsies and enabling more timely interventions.