AI can use preoperative mammograms to predict postoperative breast cancer recurrence, a 2026 retrospective study has found.
Scores from a commercial AI system for mammographic breast cancer detection and diagnosis exhibited the same predictive performance as existing clinical risk models in assessing the risk of cancer recurrence after ductal carcinoma in situ (DCIS) treatment.
AI in Breast Radiology
The general use of AI in breast imaging is not without contention. A 2025 study found that nearly one in three cancers might be overlooked by current AI tools, particularly in dense breast tissue and in patients with tumours sized 2 cm or smaller.
Mammographic AI systems are increasingly investigated for the potential they hold in breast cancer risk prediction. Emerging research has also found that AI can improve diagnostic accuracy and specificity in breast MRI.
DCIS
One in five new diagnoses of breast cancer found through screening are DCIS, where cancer cells are found in the milk ducts. Many patients with DCIS have no symptoms and are at a higher risk of developing invasive breast cancer and mortality.
Around nine in 1,000 women diagnosed with DCIS develop invasive breast cancer each year.
Predictive Power of AI
The five-centre study included more than 1,700 patients, with an average age of 55, who underwent surgery for DCIS between January 2012 and December 2017 and had at least one year of postoperative follow-up.
Medical records were analysed to identify second breast cancers. The commercial AI system for breast cancer detection and diagnosis processed preoperative mammograms.
Post-breast-conserving surgery (BCS) ipsilateral recurrence was recorded in 28 women, seven developed post-mastectomy ipsilateral recurrence, and 25 developed contralateral breast cancer.
An AI score of 73.5% or more showed significant independent association with post-BCS ipsilateral recurrence at 5 years and 10 years.
Predictions of post-BCS ipsilateral recurrence were not significantly different between AI scores and existing clinical models (VNPI and MSKCC nomogram) at 5 years and 10 years.
The Future of AI for DCIS
Researchers emphasised that AI scores, noninvasively obtained from preoperative mammography, might help guide DCIS treatment and surveillance strategies.
References
Yoon JH et al. Commercially available artificial intelligence score on preoperative mammography for prediction of future breast cancer after DCIS treatment. AJR. 2026;DOI:10.2214/AJR.25.34364.
National Health Service (NHS) England. Ductal carcinoma in situ (DCIS) data story. 2023. Available at: https://digital.nhs.uk/ndrs/data/data-stories/ductal-carcinoma-in-situ. Last accessed 18 February 2026.
