EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Potential Time and Recall Benefits for Adaptive AI ‐Based Breast Cancer MRI Screening

Luuk Balkenende iD, Jonatan Ferm, Vivien van Veldhuizen, Joren Brunekreef, Jonas Teuwen, Ritse M Mann iD

Journal of Magnetic Resonance Imaging · 2025

DOI: 10.1002/jmri.70034 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Magnetic Resonance Imaging, 2025. Verfasst von Luuk Balkenende und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Radiology, Netherlands Cancer Institute (NKI), Amsterdam, the Netherlands. Hauptthemen laut MEDLINE: Breast Neoplasms, Magnetic Resonance Imaging, Artificial Intelligence, Early Detection of Cancer. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: KWF Kankerbestrijding.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Abbreviated breast MRI protocols are advocated for breast screening as they limit acquisition duration and increase resource availability. However, radiologists' specificity may be slightly lowered when only such short protocols are evaluated. An adaptive approach, where a full protocol is performed only when abnormalities are detected by artificial intelligence (AI)‐based models in the abbreviated protocol, might improve and speed up MRI screening. This study explores the potential benefits of such an approach.

Purpose

To assess the potential impact of adaptive breast MRI scanning based on AI detection of malignancies. Study Type Mathematical model. Field Strength/Sequence Breast cancer screening protocols. Assessment Theoretical upper and lower limits on expected protocol duration and recall rate were determined for the adaptive approach, and the influence of the AI model and radiologists' performance metrics on these limits was assessed, under the assumption that any finding on the abbreviated protocol would, in an ideal follow‐up scenario, prompt a second MRI with the full protocol. Statistical Tests Estimated most likely scenario.

Results

Theoretical limits for the proposed adaptive AI‐based MRI breast cancer screening showed that the recall rates of the abbreviated and full screening protocols always constrained the recall rate. These abbreviated and full protocols did not fully constrain the expected protocol duration, and an adaptive protocol's expected duration could thus be shorter than the abbreviated protocol duration. Specificity, either from AI models or radiologists, has the largest effect on the theoretical limits. In the most likely scenario, the adaptive protocol achieved an expected protocol duration reduction of ~47%–60% compared with the full protocol. Data Conclusion The proposed adaptive approach may offer a reduction in expected protocol duration compared with the use of the full protocol alone, and a lower recall rate relative to an abbreviated‐only approach could be achieved. Optimal performance was observed when AI models emulated radiologists' decision‐making behavior, rather than focusing solely on near‐perfect malignancy detection. Evidence Level Not applicable. Technical Efficacy Stage 6.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Breast NeoplasmsMagnetic Resonance ImagingArtificial IntelligenceEarly Detection of CancerHumansFemaleBreastSensitivity and SpecificityImage Interpretation, Computer-AssistedAlgorithmsTime FactorsModels, TheoreticalReproducibility of Results

Stichwörter der Autor:innen

artificial intelligencebreast cancer screeningexpected protocol durationrecall ratescreening protocoltheoretical limits

Autor:innen und Institutionen

  • Department of Radiology, Netherlands Cancer Institute (NKI), Amsterdam, the Netherlands.
  • Department of Medical Imaging, Radboud University Medical Center, Nijmegen, the Netherlands.
  • Department of Radiation Oncology, Netherlands Cancer Institute (NKI), Amsterdam, the Netherlands.
  • Faculty of Science, University of Amsterdam, Amsterdam, the Netherlands.

ORCID-Profile: Luuk Balkenende, Ritse M Mann

Förderung

  • KWF Kankerbestrijding

Zitieren

APA

Balkenende, L., Ferm, J., van Veldhuizen, V., Brunekreef, J., Teuwen, J., & Mann, R. M. (2025). Potential Time and Recall Benefits for Adaptive AI ‐Based Breast Cancer MRI Screening. Journal of Magnetic Resonance Imaging. https://doi.org/10.1002/jmri.70034

Vancouver

Balkenende L, Ferm J, van Veldhuizen V, Brunekreef J, Teuwen J, Mann RM. Potential Time and Recall Benefits for Adaptive AI ‐Based Breast Cancer MRI Screening. Journal of Magnetic Resonance Imaging. 2025. doi:10.1002/jmri.70034

BibTeX
@article{Balkenende2025,
  author = {Balkenende, Luuk and Ferm, Jonatan and van Veldhuizen, Vivien and Brunekreef, Joren and Teuwen, Jonas and Mann, Ritse M},
  title = {Potential Time and Recall Benefits for Adaptive AI ‐Based Breast Cancer MRI Screening},
  journal = {Journal of Magnetic Resonance Imaging},
  year = {2025},
  doi = {10.1002/jmri.70034}
}
Kostenlos starten →