Clinical Scorecard: Effects of Virtual Dynamic Contrast-Enhanced Breast MRI on Lesion Size, BI-RADS Assessment, and Reader Confidence
At a Glance
Category
Detail
Condition
Breast lesion assessment using virtual dynamic contrast-enhanced (vDCE) MRI
Key Mechanisms
Deep learning generates virtual counterparts of dynamic contrast-enhanced MRI timepoints from non-contrast breast MRI sequences.
Target Population
Women undergoing clinically indicated breast MRI; the retrospective study included 540 women at one centre and a secondary test set of 52 consecutive examinations.
Care Setting
Clinical breast MRI at University Hospital Erlangen for preoperative assessment, high-risk screening, evaluation for recurrence, and assessment of indeterminate findings.
Key Highlights
The study evaluated whether a previously described vDCE approach transferred to data acquired on a scanner not used for model training.
The study examined ACR BI-RADS-based assessment in comparison with multiparametric DCE-MRI.
The retrospective cohort included 540 women imaged on one 3-T scanner and was randomly partitioned into training (377), validation (81), and holdout test (82) sets.
A secondary test set comprised 52 consecutive examinations acquired on another 3-T scanner using a matching protocol.
The excerpt describes a multiparametric protocol including T2-weighted, diffusion-weighted, and T1-weighted DCE imaging; it does not specify the DWI b-values or DCE acquisition timing.
Guideline-Based Recommendations
Diagnosis
The study evaluated ACR BI-RADS-based assessment in comparison with multiparametric DCE-MRI.
Management
The article background states that non-contrast techniques may be considered for selected clinical indications and are not proposed as universal substitutes for GBCA-enhanced MRI.
Monitoring & Follow-up
Not reported in the supplied excerpt.
Risks
The article background notes costs associated with GBCA administration and research concerns that medically used gadolinium may enter and contaminate surface waters.
Patients with breast implants were excluded because distortion artefacts frequently occur on DWI.
Patient & Prescribing Data
Retrospective study of women receiving clinically indicated breast MRI from January 2017 through June 2020; mean age was 52 years (SD 12).
The described DCE protocol used intravenous gadobutrol at 0.1 mmol/kg body weight; the supplied article excerpt does not report study results or prescribing recommendations.
Clinical Best Practices
Study protocol and eligibility details, rather than clinical best practices: the excerpt states that cases without histopathological reference data were excluded from the test sets.
The excerpt also describes exclusions involving histopathologically confirmed cancer treated with neoadjuvant chemotherapy when clinical reports documented no residual findings.
by Andrzej Liebert, Hannes Schreiter, Lorenz A. Kapsner, Jessica Eberle, Chris M. Ehring, Dominique Hadler, Sabine Ohlmeyer, Frederik B. Laun, Ramona Erber, Julius Emons, Michael Uder, Evelyn Wenkel, Sebastian Bickelhaupt
Breast cancer management is becoming increasingly biology-driven, with treatment decisions extending well beyond anatomic stage and conventional receptor classification.
The evolution of breast cancer care is increasingly challenging a long-standing assumption in oncology: that greater treatment intensity necessarily translates into better outcomes.
