Research summary ·
Artificial Intelligence in Breast Cancer Diagnosis, Treatment, and Perioperative Management: A Systematic Review of Current Applications and Future Perspectives.
Written based on 1 research study on PubMed, searched on 3 Nov 2026.
What was studied
This review looked at artificial intelligence in breast cancer care. It covered screening and imaging. It covered pathology slides and radiomics too. It also covered care around surgery, outcome prediction and language tools. It kept only original studies. They had patient-level or image-level data. The team rated bias with QUADAS-2 and PROBAST. 17 studies met the criteria from 5,275 records.
What they found
In screening mammography, deep learning gave AUC values from 0.876 to 0.939. One randomised trial found it not worse than double reading. In pathology, AUC values ran from 0.985 to 0.994. It beat unassisted pathologists at finding lymph node spread. MRI radiomics predicted complete response in HER2-positive cancer. The AUC was 0.80. No study of care around breast surgery met the criteria.
What it means for patients
The authors say AI showed clinically meaningful performance. This was so in areas with qualifying studies. They say the lack of studies around surgery is the largest gap. They say predicting receptor status without immunohistochemistry needs safety checks first.
Why it is not certain yet
The authors name weak spots in the evidence. These include outside validation, diversity of people and use in real care. Only 17 studies met the criteria. Small radiomics studies had a high risk of bias. Few imaging studies reported results by breast density.
Source
Artificial Intelligence in Breast Cancer Diagnosis, Treatment, and Perioperative Management: A Systematic Review of Current Applications and Future Perspectives.
PubMed · PMID 42922241We restate facts from the source record in our own words. The original wording belongs to its publisher.
This page is not medical advice. Talk to your doctor about your own care.
Back to Metastatic breast cancer publications