New research suggests the potential of employing artificial intelligence (AI) in real-time to help determine whether to proceed with a full breast magnetic resonance imaging (MRI) examination or opt for an abbreviated breast MRI (AB-MRI).
For the retrospective study, recently published in Radiology, researchers performed a simulation study with a cohort of 863 women to assess the use of an AI tool for facilitating a decision to continue with a full breast MRI or decide whether an abbreviated MRI was sufficient. The study authors said the AI tool generated a suspicion score for malignancy based on subtraction maximum intensity projection MRI scans.
In comparison to the full MRI protocol, the researchers found that the AB-MRI protocol offered comparable sensitivity (88.2 percent vs. 86.3 percent) and specificity (80.8 percent vs. 81.4 percent).
Abbreviated MRI also yielded similar positive predictive value (PPV3) for obtained biopsies (23.6 percent vs. 24.7 percent) and cancer detection rate (CDR) per 1,000 exams (31.6 vs. 30.9) in contrast to full breast MRI, according to the study authors.
“This reader simulation study demonstrated that image-based artificial intelligence (AI) triage has the potential to direct 25% or even 50% of examinations to abbreviated breast MRI (AB-MRI) while achieving similar diagnostic performance to conventional scanning with the full MRI protocol. In this study, there were no AI-triaged examinations for which adding MRI sequences (ie, the full MRI protocol) would have resulted in additional cancer detection,” wrote lead study author Sarah Eskreis-Winkler, M.D., who is affiliated with the Department of Radiology at Memorial Sloan Kettering Cancer Center in New York, N.Y., and colleagues.
The study authors asserted that average breast MRI scan times could decrease by up to 33 percent with a potential shift to AB-MRI for 50 percent of patients receiving breast MRI exams.
“Shorter scan times decrease medical costs, which could increase MRI access for more patients, improve patient satisfaction, and lead to better image quality since patients are less likely to move in the scanner during shorter examinations,” added Eskreis-Winkler and colleagues.
Emphasizing that the study findings could “herald a new era of AI-directed MRI scanning,” the researchers suggested this may be a harbinger of what is to come with the potential of AI to facilitate a more personalized approach to breast MRI.
“This personalization could involve AI-directed selection of MRI sequence parameters (eg, echo time and voxel size) or even real-time optimization of the k-space trajectory. Ultimately, AI could enable dynamic medical tests that acquire only the necessary and sufficient data to answer a specific clinical question for a particular patient, thus avoiding the time, cost, and exposure of acquiring extraneous data,” posited Eskreis-Winkler and colleagues.
Three Key Takeaways
1. AI-directed triage can maintain diagnostic accuracy. Image-based AI triage can effectively direct 25–50 percent of breast MRI exams to abbreviated MRI (AB-MRI) while maintaining diagnostic performance comparable to full breast MRI, with no missed cancers in AI-triaged cases, according to the study authors.
2. Efficiency gains and broader access. Shifting eligible patients to AB-MRI could reduce average scan times by up to 33 percent, potentially lowering costs, increasing access, improving patient comfort, and enhancing image quality by minimizing motion artifacts.
3. A paradigm shift toward personalized, adaptive imaging? The study highlights a shift toward AI-enabled, personalized imaging workflows that optimize acquisition parameters in real time, suggesting a future in which radiologists may supervise rather than have primary control over the acquisition and interpretation of images.
In an accompanying editorial, Fredrik Strand, M.D., praised the work of the study authors in demonstrating how AI-enabled suspicion scores may enhance triage efficiency without compromising the accuracy of breast MRI exams. However, Dr. Strand cautioned that the study may raise larger questions about the impact of AI integration upon radiologists.
“This shift — moving AI upstream into the scanning process — represents a conceptual advance toward adaptive imaging. It also raises questions about the future role of the radiologist: Will radiologists ultimately act as supervisors of AI-driven workflows, or will they retain primary control over image acquisition and interpretation? These are not merely technical questions, but also professional and ethical ones,” noted Dr. Strand, an associate professor of radiology at the Karolinska Institute in Solna, Sweden.
(Editor’s note: For related content, see “Abbreviated MRI and Contrast-Enhanced Mammography Provide Fourfold Higher Cancer Detection than Breast Ultrasound,” “Study: Abbreviated Breast MRI Offers Equivalent Accuracy to mpMRI for Women with Dense Breasts” and “Can Abbreviated MRI Have an Impact in Differentiating Intraductal Papilloma and Ductal Secretion?”)
In regard to study limitations, the study authors acknowledged that in order to evaluate the impact of triaging up to 50 percent of examinations to AB-MRI, they simulated AI-directed MRI triage for exams that were in the lower half of the AI suspected score range and acknowledged that this approach doesn’t reflect real-time decision-making in a clinical setting. The researchers also suggested that further subgroup analysis is necessary given that the cohort included pre- and post-menopausal women, people with and without a history of breast cancer, and women with breast implants.
