← All publications

conference · 2025 · Published (e-poster)

Assessment of AI Performance in Detecting Breast Lesions on MRI: A Retrospective Study in a Chinese Cohort

Siming Bayer · Adarsh Bhandary Panambur · Xiaoyang Du · Robert Grimm · Heinrich von Busch · Andreas Maier · Juan Huang · Su Lui

DOI ↗Poster ↗
01 / SUMMARY

What this paper does

A breast MRI AI prototype developed in Europe was evaluated retrospectively on an external Chinese cohort — the kind of test that decides whether a model generalizes or merely fits its development population. 200 cases were sampled from a 4,000-patient West China Hospital breast MRI cohort acquired between November 2017 and August 2023, using multi-sequence 3T MRI with expert lesion annotations. Case-level sensitivity was 0.906, with a median of 0 false positives per case, an average of 0.635, and a maximum of 7 in a single case. This work came out of a guest research stay at the hospital.

02 / METHOD

How

Retrospective sampling of 200 cases from a 4,000-patient breast MRI cohort, multi-sequence 3T acquisition with expert lesion annotations, evaluated at case level for sensitivity and per-case false-positive count.

03 / FINDINGS

Reported results

  • Case-level sensitivity 0.906.
  • False positives: median 0 per case, average 0.635 per case, maximum 7 in a single case.
  • 200 cases sampled from a 4,000-patient West China Hospital cohort, November 2017 to August 2023.
04 / WHY IT MATTERS

Relevance

For breast MRI to be usable at volume, sensitivity alone is not the constraint — the false-positive burden per case is what determines whether radiologists can work with the output.

05 / LIMITATION

What it does not show

Retrospective, single external centre, 200 sampled cases, published as a conference e-poster. Specificity was not reported in the exhibit.

06 / REFERENCE

Citation

S. Bayer, A. Bhandary Panambur, X. Du, R. Grimm, H. von Busch, A. Maier, J. Huang, S. Lui. Assessment of AI Performance in Detecting Breast Lesions on MRI: A Retrospective Study in a Chinese Cohort. ECR 2025, EPOS C-21110. doi:10.26044/ecr2025/C-21110

View original abstract

Retrospective external evaluation of a breast MRI AI research prototype on multi-sequence 3T examinations from West China Hospital, with expert lesion annotations, reporting case-level sensitivity and per-case false-positive burden.

Related work

Connected publications

BE-WISE architecture: pre-contrast, post-contrast and subtraction MRI volumes stacked into a (32, 3, 224, 224) tensor, encoded by a shared Swin-Tiny slice encoder into (32, 768) embeddings, then split into a slice-level lesion predictor producing a per-slice probability curve and a breast-level classifier producing normal, benign or malignant

Bildverarbeitung für die Medizin (BVM) 2026, 10–17

BE-WISE: Breast MRI Evaluation with Weakly-Informed Slice-Level Explanation

Published

Breast MRI models usually need lesions outlined voxel by voxel, which does not scale. BE-WISE asks how far one annotated slice per exam gets you. Pre-contrast, post-contrast and subtraction volumes are stacked and encoded slice by slice with a shared Swin-Tiny transformer. One branch predicts a per-slice lesion probability against a Gaussian target centred on the radiologist's slice; the other pools the same embeddings by attention into a normal / benign / malignant decision. On the multicenter ODELIA dataset this reached test AUC 0.8683 and an ODELIA score of 0.7098, against 0.7968 and 0.5695 for a medical slice transformer baseline.

Explainable AIMedical ImagingDetection / Localization
1 min read → Paper ↗
BE-WISER architecture: pre-contrast, post-contrast and subtraction inputs go through a SwinV2-Tiny slice encoder into slice embeddings, which split into a breast-level classifier module with ABMIL attention, feature aggregation, global head and global logits, and a weak slice-level lesion localization branch with a slice head, slice logits and per-slice class probabilities over normal, benign and malignant

Research Square preprint (in review)

BE-WISER: Class-Aware Weak Slice-Level Supervision for Breast MRI

Preprint · in review

BE-WISE could say which slice contained something; BE-WISER says what that something is. Each slice inside the Gaussian region around the radiologist's marked slice inherits the exam's class — benign or malignant — while the rest are labelled normal, so the slice head becomes three-class instead of binary. On the latest multicenter ODELIA release (926 / 230 / 292 examinations from six institutions) this reached breast-level AUC 0.864 and an ODELIA score of 0.699, against 0.807 and 0.591 for a breast-level baseline. Localization also improved: Hit@1 rose from 0.554 to 0.632 and the mean slice distance fell from 2.838 to 2.505 slices.

Explainable AIDetection / LocalizationMedical Imaging
1 min read → Paper ↗
conference 2026 Vision-Language Models · Foundation Models

European Congress of Radiology (ECR) 2026, scientific exhibit C-15780

Lesion-Aware AI for Mammography: Multi-Dataset Pretraining with ROI-Guided Contrastive Learning and Clinical Image Retrieval

Accepted

Global contrastive pretraining lets a mammography model align on breast density or view rather than on the lesion, because those cues are enough to match an image to its report. This work adds region-of-interest guidance to multi-dataset contrastive pretraining, so localized pathology has to contribute to the representation, and then tests the embeddings on clinical image retrieval — finding visually and diagnostically similar prior cases. Accepted as a scientific exhibit at ECR 2026; quantitative results are not yet publicly available.

Vision-Language ModelsFoundation ModelsMedical Imaging
Details → DOI ↗