The RSNA Conference, the annual conference of the Radiological Society of North America, held every year at Chicago’s McCormick Place Convention Center, and still the largest annual medical conference in the world, continues to evolve forward with the times. As I’ve noted in past reports, in the RSNA Conference of today is vastly different from what it was in 1990 when I began attending.
Back then, it was all about the modalities on the exhibit floor, with radiology chiefs and other radiologists being courted by vendor reps eager to sell them the latest CT, MR, and PET machines; and the educational sessions were purely clinical, meaning, about how best to consider and diagnose clinical problems. Fast-forward to the present, and both the exhibit halls and the educational sessions have been transformed; on the exhibit floor, the individuals wandering around from booth to booth are far more likely to be hospital and health system administrators than they were 35 years ago, and new-equipment purchase that’s not replacement purchase is being made relatively rare by the diminishing sizes of hospital and health system budgets. Meanwhile, the educational sessions not only are focusing on subjects never dreamed of 35 years ago, like health equity and information technology interoperability; the emergence of artificial intelligence is becoming a game-changer for practicing radiologists, and as a result, ample space is being made for AI-related discussion.
It was slightly disconcerting to see the number of AI-related sessions decline a bit this year from the volume last year, but I’m going to chalk that up to chance variation and will anticipate that the number of such sessions will increase again next year. In any case, the level of depth and breadth of the AI-related sessions was certainly impressive this year, and it’s clear that radiologists are helping to lead the way in U.S. healthcare in figuring out ways to leverage AI strategically and thoughtfully.
Indeed, what seemed clear this year is the nearly unlimited range of possibilities, clinical, clinical-operational, and operational, across the specialty. Broadly speaking, radiologist leaders are focusing on a few overarching areas: AI to support initial diagnostics; AI for clinical decision support around type of diagnostic test to order; AI to support intelligent scheduling and protocoling; the use of large language models to support patient record and history summarization; and the use of LLMs to facilitate the translation of radiology reports and information into patient-friendly language and framing.
As Arun Krishnaraj, M.D., M.P.H., a professor of radiology and medical imaging at the University of Virginia, told attendees on Tuesday in a session entitled “Improving Patient-Centered Care in Radiology Using LLMs: Opportunities and Challenges,” “Unfortunately, radiology reporting, even in the 21st century, still looks like it could be produced on a 20th-century typewriter. It’s filled with jargon and long lists.” The good news? Ai is here to rescue the situation. He and other presenters in that session described how they and their colleagues are now actively leveraging large language models to provide lay-friendly reports to patients, something that Dr. Krishnaraj and others believe will become no longer a “nice-to-have,” but instead, a necessity, as patients become empowered and take a more active part in their care in the years ahead.
And there are so many different possibilities along so many dimensions that Eric Topol, M.D., a bestselling author and a practicing cardiologist at the Scripps Clinic in San Diego and editor-in-chief of Medscape, felt confident in telling the standing-room-only audience at the plenary session on Monday, he believes that that artificial intelligence will transform the practice of medicine in the coming years.
Speaking to a standing-room-only audience at the Arie Crown Theater, Dr. Topol, author of the 2019 bestseller Deep Medicine, walked his audience of radiologists and others involved in radiology, through the evolution to date of artificial intelligence, and then predicted based on progress so far, what will happen next.
Top said that a new era in which AI tools will help physicians better diagnose and treat, and even predict the onset of, disease, is just on the horizon for U.S. healthcare. He said that the foundational work over the past numerous years in developing algorithms and working with large language models, has set the stage for massive change. For example, the data gathered from enormous amounts of data and images, is already leading to better diagnoses, as in the case of gastroenterology, where gastroenterologists are already using AI-facilitated endoscopy to achieve detect more polyps than they could previously. And data is being gathered even from such diagnostic images as x-ray, creating massive lakes of data that are being used to support physician diagnosis processes. This phenomenon he referred to as “Machine Eyes”—the collection of data that, when analyzed and poured into clinical decision support, can improve diagnostics. Amazingly now, studies are finding that the analysis based on chest x-rays can lead to the diagnoses of a surprising range of diseases, including diabetes. He cited a September 2023 study based on the analysis of 1.6 million retinal images gathered in the U.K. that produced breakthrough predictive diagnostics.
Meanwhile, Topol told his audience, what’s becoming clear is that “AI does a really good job of its text for completeness, correctness, and conciseness. AI reports are tighter, easier to understand, and more complete than reports produced by physicians.” He also made note of a couple of studies that have concluded not only that AI does a better job of diagnosis than human physicians, but two studies have found that AI alone actually does a better job of diagnosis than AI + humans. That result, though, he quickly added, is probably related to the fact that the studies were “contrived,” artificial tests, not based on actual patient care situations. It is interesting to note, though, he added, that AI appears to promote the expression of empathy among physicians.
Per all that, Dania Daye, M.D., Ph.D., associate professor of radiology at Harvard Medical School and director of the Precision Interventional and Medical Imaging lab in the Division of Vascular and Interventional Radiology at Mass General Brigham, in the session initiated by Dr. Krishnaraj, referenced an article in Radiology entitled “A Context-based Chatbot Surpasses Radiologists and Generic ChatGPT in Following the ACR Appropriateness Guidelines,” in which a study found that Chatbot provided substantial time and cost savings. She cited several other studies in the recent literature, including one that appeared in the October 5, 2023 edition of JAMA Network Open, entitled “Generative Artificial Intelligence for Chest Radiograph Interpretation in the Emergency Department,” in which the GPT-generated reports were found to be equivalent to radiologists in the ED and better than teleradiologists.
Dr. Daye cautioned that clinicians and data scientists need to move quickly to eliminate “hallucinations, bias reproduction, misinformation propagation, and lack of accountability.” But, given strong efforts in those areas, she said, the way is open to effectively leverage AI for patient care, education, and research.
The potential is vast, RSNA President Curtis P. Langlotz, M.D., Ph.D., had said in his president’s address on Sunday. Indeed, he noted, in the 1980s, it had taken four years to build a system that could analyze just a few images. “Today,” in contast, “wit the right training data, we can build a system in days that has better accuracy than anything that we built back then.” And, per that, Dr. Langlotz said, “Anyone who works with AI knows that machine intelligence is different, not better than human intelligence.”
And what seems clear is that those humans moving AI forward in radiology are being extremely thoughtful and are avoiding the temptation to try to “boil the ocean,” a temptation so often present in healthcare. Instead, they’re getting to work and rolling up their sleeves to tackle a range of practical problems; in the process, they will not only make radiologists more efficient and effective—an important goal as the healthcare system faces a growing shortage of radiologist person-power, as diagnostic imaging demand rises in our aging society—but they will also usher in a new era of patient engagement, another extremely important area for healthcare system progress.
And it’s obvious that we are now on the road with all of this, and that the next few years in radiology will witness tremendous progress in harnessing AI to improve radiology practice and healthcare delivery. And that is an exciting prospect, and one of the encouraging aspects of attending RSNA this year. Who knows what RSNA24 will be like? I can’t wait to find out.
