Radiology’s future hinges on using AI to relieve capacity while doubling down on human margin, accountability and trust to transform imaging data into patient care.
FROM THE CHAIR OF THE BOARD OF CHANCELLORS
It is truly an honor to write to the membership for the first time as Chair of the ACR Board of Chancellors. I am immensely grateful for the trust the Board has placed in me and am committed to continuing to serve this College and the broader House of Radiology over the coming two-year term. Before I share some overarching thoughts, I would like to begin by recognizing my predecessor, Alan H. Matsumoto, MD, MA, FACR, the incoming ACR President. Alan has been an extraordinary bridge-builder. His tenure was marked by tireless and deliberate outreach across the full House of Radiology, with particular attention to his IR colleagues and to the radiation oncology community. The relationships he strengthened, inside the College and across our sister societies, will continue to bear fruit. His leadership also set us on a path to re-evaluate and improve the Board’s structure and function. While that process is ongoing, his tenure leaves the organization in a better position than he found it, and I intend to build on the work he started.
Coming off twelve years of direct involvement with the ACR Commission on Informatics, I cannot help but notice that we have reached an inflection point when it comes to AI and radiology. But it is not the one which has recently grabbed attention due to certain external CEO statements: Radiology is not approaching obsolescence — quite the opposite. Demand for imaging care is stronger than ever; our workforce is tight but growing and qualitatively evolving owing to remote/hybrid work and consolidation trends. We are finally seeing at least the potential for the next generation of AI to contribute meaningfully to our work. Overall, I interpret this as another period requiring resilience and substantial adaptation from all of us: Our specialty has navigated several of those in its history.
Recent data do not support the claim that trainees are abandoning the field. Training positions are at an all-time high and fill rates robust, but the applicant pool has softened in recent years — and thus we cannot count on a thick cushion of interest forever. So, while the specialty remains competitive, we must chart a path towards a future quickly enough to meet demand without worsening burnout, delays, or threatening levels of commoditization. Work done by our own ACR Commission on Human Resources Commission and by the Harvey Neiman Health Policy Institute® predicts that demand for imaging will grow at least as fast, if not faster, than the radiology workforce. This means we cannot scale our way out of the problem with our traditional approaches alone.
Used in the right way, AI is less a substitute for radiologists than an enhancer of the care contribution we can make.
When we consider whether AI will ultimately be a useful “ally” or a threat we should consider this reality: AI is one of several levers available to increase the value and reach of what we do. In my opinion, the near-term impact will not be the wholesale replacement of radiologists. The value of these tools, especially those that help us manage reporting and EHR information, lies in helping us standardize work products, reduce avoidable variation, take low value cognitive and clerical tasks off our plate, support quantitative imaging, integrate clinical and imaging data, and surface relevant guidelines and risk assessments at the point of interpretation. Used in the right way, AI is less a substitute for radiologists than an enhancer of the care contribution we can make.
I have found it useful to think about this as the “human margin.” In the business world, margin refers to the slice of activity at the edge of what an enterprise does where value is created or lost. The human margin in radiology is the slice of our work where radiologist judgment makes the decisive difference for patient care — the part even a well-trained model or even ensembles of models cannot do alone no matter how sophisticated. As AI improves at detection, measurement, triage, drafting, and pattern recognition, the human margin is where we must lean in — such as clinical judgment, management of uncertainty, accountability for the recommendation, and trusted communication with referring clinicians and patients. Future radiologists will demonstrate their value-add to care by turning images, model outputs, prior imaging and other clinical information, personalized patient risk, and downstream consequences into trustworthy medical judgment that is at the center of decision-making in-patient care, including population health management. Treated defensively, AI narrows us to the very tasks that are most vulnerable to automation. Treated strategically, it lets us move up the value chain and strengthen our role as the physician specialty responsible for imaging intelligence in modern healthcare.
How then do we move toward that future professional identity through action today? We need to be AI literate enough to know where these tools are useful, where they fail and how to monitor them. We need to demand of our production system vendors (EHR, PACS, RIS, Voice Recognition, Communication Systems) that they become more technically and semantically integrated and interoperable and expose interfaces to the external world that allows for control by other systems as deemed necessary by the end user (to enable information flow, and future agentic workflows). We need to redesign workflows so that AI does what it is well-suited to do, while radiologists preserve and expand our role in synthesis, consultation, appropriateness, quality and governance. And we need to use AI not simply to read faster, but to contribute more meaningfully to care, with more quantitation, more standardization, and more visibility at the point of clinical decision making.
The ACR Data Science Institute® (DSI) is leading the way, hard at work on your behalf: In the past two years alone, DSI released the most comprehensive curated collection of FDA-cleared AI products and corresponding model cards, the first national quality recognition program in AI (ARCH-AI), built an ongoing community of participating sites, launched the AI registry and data service ASSESS-AI, and convened the development of the first national Practice Parameter for Imaging AI which was recently ratified by the ACR Council at its 2026 Annual Meeting. Our teams are now working on a proposal to potentially build an accreditation program on top of this foundation, collaborating with the ACR Economics and Government Relations teams on coding and payment issues, and closely communicating with the FDA and industry stakeholders on emerging challenges. There has never been a better time to become an ACR member and take advantage of these offerings while supporting these important activities!
Alongside this evolving technology frontier, a second change is reshaping how we work, and it deserves more attention than it usually gets at our meetings and in our publications. The steady migration of practice toward distributed, remote, and hybrid work arrangements has brought real benefits: Workforce flexibility, geographic access for underserved areas, sustainable arrangements that keep talented radiologists in the field, and coverage that would otherwise be unfillable. These are valuable, and presumably here to stay. The question is what doing this well requires of us professionally and of the College in supporting this new and emerging reality.
Radiology has historically built its professional “connective tissue” largely as a byproduct of physical proximity, through micro engagements in radiology’s hallways and reading rooms, mentorship, peer review, case conferences, multidisciplinary discussion, and workstation teaching and consultation on actual cases. There was a time when colleagues who interpret images and colleagues who order scans shared a physical space, and encounters were frequent and organic. A floor plan with a benefit, so to speak. In our increasingly digital and distributed practice, none of this comes for free; rather we must earn it purposefully. Without intentional design, the practice becomes a transactional arrangement in which cases flow in and reports flow out, and the radiologist becomes a contractor to a workflow rather than a member of a clinical team. This connects directly to my human margin point: The professional structures that develop and sustain clinical judgment, accountability, communication and trust are the same ones that distributed practice now requires us to construct deliberately. We can do so by leveraging modern communication tools but absolutely need to take that initiative as it represents a path to sustained relevance.
There is a third topic I wish to “plant” in this inaugural column — Advocacy. ACR continues to advocate on behalf of the entire House of Radiology, diagnostic and interventional radiologists, radiation oncologists, medical physicists, and nuclear medicine physicians. ACR can only use member dues, no other source of revenue, to support the all-important advocacy which ensures our professional future. It is more important than ever that our message be unified when we engage with federal and local legislators, regulators, agencies and payers. There has been a recent trend of fragmentation in advocacy, as some organizations and consolidated practices prefer to deliver their own message. I would remind everyone that ACR is by far the largest voice and the most impactful organization working on your behalf. In the long run, we may all be better off with one organization and one voice, even with some compromise on specific points, than with a myriad of individual messages that dilute the field’s influence. There is also a business case here. The full House of Radiology currently maintains more than 60 subspecialty associations, each with its own administrative framework, all of which need to be maintained and funded. When time and money get scarce, we may be better off together than apart. I understand that this is provocative, and surely it remains a choice individual societies and members will make for themselves. However, it is worth seeding here, and ACR leadership is certainly open to discussion.
Radiology has been through inflection points and technology-driven adaptation before — think about new cross-sectional imaging modalities, PACS, voice recognition…the complete digital transition. Each time the specialty absorbed substantial change and came out better for it. We will do so again. We have the talent, we have the science, and we have a College (with its amazing volunteer members and staff) doing the work to ready our profession for what comes next. The future of radiology is not a contest between radiologists and AI. It is a test of whether we use AI to relieve capacity pressure while doubling down on the human margin, judgment, accountability, context, and trust that turn imaging data into medical care for actual patients. I am confident we will rise to it, and I look forward to doing the work with you.
Your First Radiology Contract
Whether you’re considering a job in academics, private practice, hospital or teleradiology, you need to be familiar with the key contract provisions and why they matter.
Read more
Inside the New DICOE Survey Experience
From virtual deep dives to on-site insights, the revamped Diagnostic Imaging Center of Excellence (DICOE) program is changing how facilities demonstrate excellence.
Read more
The Long-Range Nature of Informatics
The outgoing and incoming chairs for the ACR Commission on Informatics talk leadership and collaboration to provide essential resources to ACR members.
Read more
