Thanks, RSNA23, It Was Real (And Also Quite Artificial)

Dec. 7, 2023
Even as dramatic change sweeps U.S. healthcare, radiology leaders focused strongly on one major changemaker—the arrival of artificial intelligence in radiological practice

Having attended more than 30 RSNA annual conferences now—the annual conferences put on by the Oak Brook, Illinois-based Radiological Society of North America, always at Chicago’s vast McCormick Place Convention Center—I can speak to how the annual global gathering of radiologists and everyone connected to radiology, has evolved over the years.

I remember my first RSNA, in 1991; it was a completely different event. True, the core clinical-educational sessions, involving the accurate diagnosing of conditions based on diagnostic images—are still fundamentally the same kind of phenomenon (with asterisks). But the exhibit floor? It is totally different nowadays. Back in 1991, the majority of visitors to the exhibit floor were practicing radiologists, many of them chiefs of radiology; and the main activity taking place on the exhibit floor was the displaying of the latest modalities—the CT, MR, PET, nuclear imaging, mammography, and x-ray machines that scan the human body—with the discussion being clinical and clinical-technological. And the radiologists were the key decision-makers, and were treated with great deference.

Then along came PACS (picture archiving and communications systems) systems, which revolutionized the field by eliminating film except in a tiny, tiny percentage of cases (probably less than one-hundredth of a percent, at this point), turning all those film-based images into digitized images, and allowing for greater accuracy and usability; and with it, RIS (radiology information systems) systems, which guided radiologist and radiological tech workflow. At the same time, EHRs (electronic health records) were emerging into reality. And so within the decade-and-a-half from 1990 to 2005, radiology had been transformed, and the discussions on the exhibit floor had morphed dramatically, and now, were more and more about imaging informatics—something that hadn’t even existed back in the 1980s.

Meanwhile, the macroeconomics of radiology was changing dramatically, as the U.S. healthcare system rushed closer and closer to a total cost cliff—the position it is in right now. As the Medicare actuaries warned us earlier this June, total annual U.S. healthcare spending, driven by the aging of the population and an ongoing explosion in chronic disease, even among children, is exploding wildly now, and we will be going from the current, already-mindblowing, $4.6 trillion a year in total healthcare expenditures, to $7.2 trillion by 2031, with 19.6 percent of our gross domestic product being consumed by healthcare expenses in that year. That’s a 34.2-percent increase in eight years—in other words, totally mindblowing.

And of course, radiologists are caught in the middle of the cost discussion, because diagnostic imaging is extremely expensive, and the purchasers and payers of healthcare in this country are paying ginormous sums for the individuals whose health insurance they’re paying for, to obtain diagnostic imaging services. Of course, there’s a huge debate going on about radiologist reimbursement, too. But in the midst of all of that, the costs keep going up, even as older radiologists retire, and those who remain in practice are being required to consistently improve their productivity, meaning to interpret studies faster and faster.

Into this landscape has emerged artificial intelligence (AI), a phenomenon set to transform radiology once again. And four years ago, there was great fear among many practicing radiologists that AI would actually displace them—meaning, that machines would be interpreting diagnostic images, and human beings would be excluded. Once it became clear that no such thing would happen, radiologists—arguably the most tech-friendly of all practicing physicians—switched mindsets faster than Beyoncé and Taylor Swift can churn out new pop-music hits—and became enthusiastic about the possibility of AI helping them.

And so that’s where we are now, and that was obvious everywhere at RSNA23, held last week at McCormick Place (Nov. 26-30). There were numerous dozens of sessions devoted to AI, all the way from the policy-related plenary sessions to very granular clinical sessions in which practicing radiologists who are already deep in the weeds on developing algorithms or working with generative AI, shared their learnings thus far on the journey. Among the latter type of sessions was Monday’s first plenary address, given by Elizabeth S. Burnside, M.D., M.P.H., senior associate dean in the School of Medicine and Public Health at the University of Wisconsin-Madison, and deputy director of the Institute of Clinical Translational Science for Breast Imaging, at the University of Wisconsin. Dr. Burnside delivered a terrific speech, looking at the challenges on every level, from policy to operational to clinical, and stating that, when it comes to ethics around algorithm development, “Policies really are part of the key,” she said. “And, we need to work diligently on developing understanding,” with the need to find the resources and support to develop data sets, and the identification of known local environments in which the tools can be tested, being important as well. “Leadership is really sitting in your seat!” she told the audience, meaning that they, the audience members must be leaders in this work. “You have an important role to play,” she concluded. “Proudly tackle the tame, while always keeping an eye on the wicked.”

Meanwhile, in a session on Tuesday entitled “Best Practices for Continuous AI Model Evaluation,” Matthew Preston Lundgren, M.D., M.P.H., a practicing radiologist and the CMIO at Nuance, emphasized how important the practical aspects of algorithm development are, with governance and ongoing management being huge elements in the ultimate success of AI development in radiology, and discussing the “Day 2 Problem,” as algorithmic models can drift and lose their effectiveness. In other words, the full spectrum of challenges and opportunities was addressed at the conference.

So it was a very, very interesting RSNA indeed. And what seemed clear is that this specialty-wide plunge into AI and machine learning will bear fruit in a number of areas—some purely clinical, but others around study prioritization and results reporting processes, of course, and also around clinical quality assurance. I particularly liked Burnside’s invocation not to give up in despair around the “wicked problem of biomedical AI development, but instead to commit to engaging stakeholders, to maintain rigor around the analysis of both quantitative and qualitative techniques; and to guide forward decision-making that is continuously aligned among stakeholders and focused on outcomes.

Every year at RSNA, there is a mix present of various psychological winds, from the robust straight-ahead optimism of some industry leaders and vendors, to Chicken Little-level panic over one or more issues. But RSNA23 convinced me that, even as the radiology field faces ginormous challenges of all kinds going forward—policy, payment, staffing, etc.—there are exceptionally brilliant people in the specialty, both clinicians and non-clinicians—who are going to help us all solve problems going forward. In other words, as the French would say, “Plus ça change, plus c'est la même chose”—the more things change, the more they stay the same.

So, farewell, RSNA23, it’s been real—and also very artificial (intelligence). I look forward to experiencing the Zeitgeist at RSNA24, when the conference once again returns to Chicago’s McCormick Place during the week after Thanksgiving.

 

 

Sponsored Recommendations

How to Build Trust in AI: The Data Leaders’ Playbook

This eBook strives to provide data leaders like you with a comprehensive understanding of the urgent need to deliver high-quality data to your business. It also reviews key strategies...

Quantifying the Value of a 360-Degree view of Healthcare Consumers

To create consistency in how consumers are viewed and treated no matter where they transact, healthcare organizations must have a 360° view based on a trusted consumer profile...

Elevating Clinical Performance and Financial Outcomes with Virtual Care Management

Transform healthcare delivery with Virtual Care Management (VCM) solutions, enabling proactive, continuous patient engagement to close care gaps, improve outcomes, and boost operational...

Examining AI Adoption + ROI in Healthcare Payments

Maximize healthcare payments with AI - today + tomorrow