If it’s time for the RSNA Conference, it’s time to check in again with one of the industry’s leading imaging informatics experts, Joe Marion, principal in the Waukesha, Wisconsin-based Healthcare Integration Strategies consulting firm.
On Tuesday, December 3, as the annual RSNA Conference, sponsored by the Oak Brook, Illinois-based Radiological Society of North America, was rolling forward at Chicago’s vast McCormick Place Convention Center, Marion, who was attending his forty-fourth RSNA Conference, sat down with Healthcare Innovation Editor-in-Chief Mark Hagland to converse about this year’s conference, and the underlying trends taking place out in the industry right now.
What are you seeing this year that’s different from in previous years?
AI [artificial intelligence] is obviously the buzzword this year, and it’s everywhere—but is AI actually any further along and real this year, out in the industry? Maybe some applications are being implemented here and there; but it’s still in its infancy, and it’s not progressing rapidly. Customers—providers—are asking themselves, how can I make money with this, when my reimbursement continues to get squeezed? So there’s a trust factor. They’re asking, is it going to make me that much more efficient? So the jury’s still out. And the vendors are taking this posture of, they’re hedging their bets in the context of saying, if there’s a third-party solution, we can plug it in; if a provider wants to do some of their own research, I’ll give them the API to plug that in. So I think they’re hedging their bets.
And they’re rebranding everything; GE is rebranding everything as “Edison,” and they’re embracing AI as a part of that. Some of the others are not quite so aggressive; Sectra is taking a more conservative approach; they’ll do it, they’ll use third-party vendors. But they won’t call it a platform. Everyone’s talking enterprise imaging. And the question is, what’s the state of enterprise imaging? How real is it?
With regard to several presentations I’ve seen by radiologists, examples are abounding around the leveraging of AI to improve radiologist workflow and study prioritization.
Yes, that’s right. There are really two phenomena taking place right now. What you’ve just mentioned—leveraging AI in order to improve radiological workflow process and prioritize important clinical phenomena—that’s one track, and that’s moving forward in place. The broader use of AI that’s been predicted for years—the use of algorithms for diagnosis—that is still lagging at this point; and that’s the use that had been widely predicted.
Yes, that first use that you mentioned, is really about the development of smart worklists, and a more intuitive way of directing studies for reads. There’s more potential in that area, early on, than in investing in algorithms for diagnosis.
How do you see those two tracks evolving forward?
The worklist- and efficiency-oriented case studies, smart worklists, that’s very real. The algorithms, it’s a wait-and-see, and where do you get the justification or it? Once you do get the justification, all the big vendors are talking about building platforms in which they can plug those directly in. The future will be that as you get the justification for it, it’s an add-in, if you will. It’s as though you have an application to enhance Microsoft Office, an add-in. I think that’s how most of it will play out, but it will depend on how that technology will play out economically. There’s the simplified version that simply points something out; CAD [computer-aided design] has been doing the equivalent of lighting highlighted circles around areas on a scan, for years. But if it involves the system doing some sort of analysis, such as pneumothorax or brain bleeds—the value-add will be being able to squeeze or reduce the cycle time for treating serious illnesses. That’s where I think the economic analysis will play out.
What does the exhibit floor look like to you?
More of the same, essentially. But the other element in this is the increase in the use of streaming technology. In the past, if I were a clinician, I would have had a PACS [picture archiving and communications system] archive, and a workstation, and I would pick a set of images and the server would send them. With the new technology, everything resides on the server; and because of the speed of networks, and everything—what happened to Blockbuster? You used to go rent a CD or whatever. It’s all streaming now. And they’ve reached the point where I can stream a movie from wherever. The same is happening with diagnostic workstations now: the data simply sits on the server. If I want to do a 3D analysis, I’m simply working with a viewer tool. So if I want to read from home or in the doc-in-a-box or wherever, it doesn’t matter where the images are, I can just read live, streamed. And I think the crucial developments will be the ability to stream to the patient—no need to burn a CD. So providers can simply allow patients to view their images, without any data being actually sent to them. They’ll just be viewing.
In the emergency medical services context, will streaming make it easier for all providers?
Yes, exactly, because it’s immediately available at all locations. About seven years ago, I had an episode of vertigo, and they wanted to do an MRI, because of enlarged ventricles in my brain (where the cerebral spinal fluid is). And I had a whole record that explained my condition. But had I been able to say, go to this link, and you’ll see my MRI from 2006, you’ll be able to see the images. But I couldn’t do that. It was all on CDs or in my local PACS. So that’s the promise of the future.
Imagine that I see a primary care physician, and they want me to see a specialist; what happens if they want me to see a specialist out of network, what then? With streaming, it’s no problem. And at Visage, which has Partners HealthCare and Mayo Clinic as customer-partners, they were among the first to make streaming go live. They went live at least five years ago. Fuji jumped on the bandwagon a couple of years ago. Carestream was moving in that direction; now, since Philips acquired the IT portion of Carestream, that’s a part of that portfolio.
So we won’t need to be sending heavy sets of images hither and yon, via HIEs, then, correct?
Exactly. And if I look at the Milwaukee market, Aurora, Froedtert, and Ascension have all converged onto Epic. And from a medical record perspective, even though my provider is Aurora, they can go back and get information from my previous provider via Epic. The same thing could happen with images.
You’ve been pondering the shift towards enterprise imaging, lately. Where is that concept right now, in the industry?
So, enterprise imaging encompasses workflow orchestration and clinical decision support; the question is, will everything be on separate platforms? Will the VNA [vendor-neutral archive], image viewer, workflow orchestration/clinical decision support, AI, all be connected on via some kind of enterprise imaging-based architecture?
That’s the technical piece. The business piece of it is, where’s the motivation, where’s the driver to do all this? What’s the business case? I look at Aurora. They have no capability in this area. Why? It’s not included in their priority list. They did PACS, the VNA, and they’re focused on Epic as their EHR [electronic health record]. From the patient portal side, there’s a huge need for this, as patients become more informed. So I want to get the perspectives of additional providers, as I further research this. And the vendors are wrestling with it, because nobody’s coming to them saying, I’ve got to have all of this. And the question is, who is the driver within the patient care organization? Is it the chief medical officer? It certainly won’t be the chief of a particular specialty. Is it the practicing specialist? We don’t know yet who will drive this work. Somebody said to me today, you’re the evangelist of enterprise imaging. That’s what intrigues me; how do you get people excited about doing this? But as with other types of technologies, the trend tends to start at academic medical centers.
What would your advice be for CIOs, CMIOs, and other senior healthcare IT executives, right now, in this landscape?
They need to find out what their roadmap will be at the enterprise level. Instead of letting all the individual service lines do their own things, they need to develop an integrated vision. The University of Michigan. A senior leader who was there is now at Spectrum Health in Michigan.
So they need to think long term about what their enterprise strategy is?
Yes. And they need to find a change agent. Who else I the organization has that broad perspective, and a vested interest? They have to create the governance organization to make this happen within the organization. AI won’t just be a radiology thing; it will be cardiology, pathology, every specialty. It can’t just be a proliferation of solutions; they need to have some kind of strategy.
