Enhancing Early Detection of Contralateral Breast Cancer in Breast Cancer Survivors Using AI-Assisted Mammography.
Authors
Affiliations (8)
Affiliations (8)
- Department of Surgery, Seoul National University College of Medicine, Seoul, Republic of Korea.
- Genomic Medicine Institute, Seoul National University Medical Research Center, Seoul, Republic of Korea.
- Cancer Research Institute, Seoul National University, Seoul, Republic of Korea.
- Department of Radiology, Seoul National University College of Medicine, Seoul, Republic of Korea.
- Department of Radiology, Seoul National University College of Medicine, Seoul, Republic of Korea. [email protected].
- Department of Surgery, Seoul National University College of Medicine, Seoul, Republic of Korea. [email protected].
- Genomic Medicine Institute, Seoul National University Medical Research Center, Seoul, Republic of Korea. [email protected].
- Cancer Research Institute, Seoul National University, Seoul, Republic of Korea. [email protected].
Abstract
Women with a history of breast cancer face an elevated risk of developing contralateral breast cancer (CBC). Although annual mammographic screening is recommended, its sensitivity is limited, particularly in younger women and those with dense breast tissue. This study assessed the potential of artificial intelligence (AI)-based computer-aided diagnosis (AI-CAD) to improve CBC detection in a cohort of breast cancer survivors and to enhance early CBC detection. This study included 454 women who developed CBC and 454 matched controls without recurrence. Mammograms were analyzed using AI-CAD software assigning abnormality scores from 0 to 100, with scores over 10 indicating CBC. Standalone AI predictions were compared with radiologists' initial assessments, and its ability to detect early signs of CBC in prior mammograms was evaluated. The median age of patients was 53 years; the majority (79.9%) had dense breast tissue. Standalone AI detected 271 CBC cases solely from mammography, achieving 6.2% higher sensitivity compared with radiologists (59.7% vs. 53.5%, pā=ā0.009). Notably, AI detected 66 CBC cases (14.5%) that radiologists had missed on mammographic assessment alone. With respect to earlier detection, 81 cases (29.9%) were identified on prior mammograms at a median of 13.3 (9.6-19.9) months before clinical diagnosis. Of these, 28 cases (10.3%) were detected more than 6 months before pathological confirmation, and 53 cases (19.6%) more than 1 year earlier. AI-CAD may enhance the detection of CBC in breast cancer survivors and facilitate earlier identification on surveillance mammography. Further studies are needed to assess its integration into surveillance mammography and clinical impact.