UW Radiology contributes to international scientific statement on AI in cardiac imaging

A headshot of Dr. Domenico Mastrodicasa. In the photo, Dr. Mastrodicasa wears a dark suit coat, white button-up shirt, and red tie. He has dark hair and a broad smile.

UW Radiology Acting Instructor Domenico Mastrodicasa, MD, has published a scientific statement on artificial intelligence (AI) in cardiac CT and MRI in the journal Radiology. The statement is titled, “Use of AI in Cardiac CT and MRI: A Scientific Statement from the ESCR, EuSoMII, NASCI, SCCT, SCMR, SIIM, and RSNA.”

In the scientific statement, Dr. Mastrodicasa and colleagues explored where AI is already making an impact in the cardiac imaging workflow, how ready these tools really are for clinical use, and what challenges still need to be addressed. As part of this effort, Dr. Mastrodicasa and colleagues adapted NASA’s Technology Readiness Level (TRL) scale, originally developed for space technology, to assess AI maturity in cardiac imaging. This scale looks like:

  • TRL 1–2: Research phase (exciting ideas, but early days)
  • TRL 3–6: Development (AI tools being tested and refined)
  • TRL 7–9: Clinical deployment (ready, or almost ready, for real-world use)
Figure 1 - A circular infographic illustrates the role of artificial intelligence (AI) across the cardiac imaging workflow, with a stylized 3D heart at the center, symbolizing AI’s integration in cardiovascular imaging. The infographic is divided into eight labeled segments, each representing a different stage of the imaging process: Patient & Imaging Test Selection Protocolling & Workflow Optimization Image Acquisition & Reconstruction Image Analysis & Interpretation Detection of Incidental Findings Risk Stratification & Prognostication Imaging Report AI Deployment Additionally, "Ethics, Legal, and Environmental Sustainability" is noted as a key overarching consideration. The outer background features two stylized MRI scanners, reinforcing the clinical imaging context. The diagram visually summarizes how AI is applied at every stage, from test selection to radiology reporting and AI deployment.

Figure 1

Figure 3 - A color-coded bar chart illustrates the technology readiness levels (TRLs) for artificial intelligence (AI) tools in cardiac CT and MRI, ranging from early research (level 1) to full clinical deployment (level 9). The background transitions from red (low readiness) to green (high readiness), visually representing progress toward clinical adoption. On the left, research-phase AI applications (TRLs 1–3) include patient and imaging test selection, cardiac imaging workflow optimization, and scanning protocol selection. In the middle, development-phase AI tools (TRLs 4–6) involve image acquisition, reconstruction, and analysis for cardiac and non-cardiac findings. Some emerging applications, such as AI prognostication for major adverse cardiac events and large language models for cardiac imaging reports, are shown at an intermediate readiness level. On the right, clinically deployed AI tools (TRLs 7–9) are highlighted in green with corresponding circular images, including deep learning-based coronary CT angiography reconstruction, CT-derived fractional flow reserve, plaque analysis, cardiac MRI biventricular function assessment, and incidental coronary artery calcium (CAC) detection. The figure provides a visual summary of AI’s progress in cardiovascular imaging from research to clinical implementation.

Figure 3

By evaluating AI through this scale, this work bridges the gap between innovation and clinical implementation. Beyond assessing AI readiness, the statement takes a holistic approach, addressing key considerations such as ethics, fairness, transparency, bias mitigation, and sustainability. Key takeaways and recommendations grounded in the collective expertise of the team of authors are summarized in Figure 4.

Figure 4 - A stylized digital illustration of a human heart, glowing in red and blue, sits atop a computer chip, symbolizing the integration of artificial intelligence (AI) in cardiac imaging. To the right, a structured text box titled "Key Recommendations and Considerations" outlines insights on AI adoption in cardiac CT and MRI, including its potential benefits, risks such as hallucinations, the importance of diverse data sources for validation, and the need for independent cardiac imager training to develop independent diagnostic skills without overreliance on AI. Below, the figure legend attributes the scientific statement to multiple radiology and cardiovascular imaging societies and notes that the image was generated using DALL-E (version 3.0, OpenAI).

Figure 4

The statement generated strong interest, with over 2,300 downloads in the first two weeks and an Altmetric score of 60, ranking among the top 5% of all research outputs tracked by Altmetric. Radiology Editor-in-Chief Linda Moy praised the article, recognizing its contribution to the growing conversation on AI in cardiac imaging. Engagement has also been particularly strong on LinkedIn, where the post has reached nearly 6,000 members and generated over 9,500 impressions. Following its publication, the statement has been featured in two AuntMinnie.com articles: RSNA and collaborators release statement on AI for cardiac CT, MRI and International collaboration leads to statement on AI for cardiac CT, MRI.

Dr. Mastrodicasa and co-authors hope this statement serves as a guiding framework to help cardiac imagers, AI experts, and policymakers work together to advance AI in clinical practice responsibly and effectively.

This work is the result of a multi-institutional collaboration among experts from leading institutions, including Marly van Assen, PhD (co-first author), from Emory University, Atlanta, Georgia; Merel Huisman, MD, PhD, from Radboud University Medical Center, the Netherlands; Tim Leiner MD, PhD, and Eric Williamson MD, from Mayo Clinic, Rochester, Minnesota; Edward D. Nicol MD, MBA, from the Royal Brompton Hospital, London, United Kingdom; Bradley Allen, MD, MS, from Northwestern University Feinberg School of Medicine, Chicago, Illinois; Luca Saba, MD, from University of Cagliari, Cagliari, Italy; Rozemarijn Vliegenthart, MD, PhD, (co-senior author) from University Medical Center Groningen, the Netherlands; and Kate Hanneman, MD, MPH, from University of Toronto, Ontario, Canada.

Listen to Dr. Mastrodicasa discuss this work in a podcast produced by Radiology entitled Radiology & AI: The Future of Cardiac Care.

Share this story: