Early Detection of Breast Cancer in MRI Using AI.
Source check: Source checked. Not independent scientific confirmation.
Finding: A patient-level evaluated CNN predicts subsequent cancer risk from current MRI in a retrospective high-risk cohort. It is a single-site proof of concept with modest discrimination and low positive predictive value; the earliest located result-bearing version is the 2023 preprint.
Technology: A pretrained 2D CNN is fine-tuned using per-slice probabilities to predict one-year future cancer risk from current MRI; patient-level 10-fold evaluation avoids patient overlap between train/test.
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- A pretrained 2D CNN is fine-tuned using per-slice probabilities to predict one-year future cancer risk from current MRI; patient-level 10-fold evaluation avoids patient overlap between train/test.
- Test and result
- The retrospective 910-woman cohort (3,029 exams, 5,128 breasts, 115 later cancers) yielded reported AUC 0.72 and a top-risk selection with 35/115 cancers at 30% sensitivity; study is described as proof of concept.
- Limitation / still unknown
- Internal patient-level cross-validation with later cancer ascertainment; single-site retrospective cohort and low PPV/high false-discovery burden constrain use.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://doi.org/10.1016/j.acra.2024.10.014
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 3/5 · Validation 2/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.