October 17, 2016
PACS 3.0: The Next Iteration of Radiology’s Reading Platform
Oct 17, 2016 | Michael Bassett
The consensus is in: Image management strategy decisions that are found in most hospitals and practices today will be very different as we head into the future.
“In talking to other CIOs in outpatient radiology businesses and hospitals, there seems to be a migration away from PACS as a system, or at least as the PACS moniker,” says Todd Thomas, chief information officer, Austin Radiological Association (ARA), a radiology practice that operates 17 outpatient imaging centers in the Austin and Central Texas area, and which provides radiology services to 20 area hospitals. “I’ll hear hospital CIOs say they are deconstructing their PACS—they’ve got a VNA [vendor neutral archive], they’ll get a viewer, and they’ll acquire a worklist engine. But it seems the VNA is always the keystone to these conversations.
Thomas notes that ARA’s PACS provider acquired a VNA provider. “I think these PACS vendors are realizing that organizations are beginning to move away from that classic, core model of the storage, workflow and viewer all rolled into a single application,” he says.
“It will be interesting to see how many other VNA vendors get acquired over the next few years,” he adds. “I think there will be a lot more flexibility going on with this ‘best of breed’ approach, rather than with the classic sort of PACS 1.0 or 2.0 model.”
However PACS evolves as we head into the future, there’s no question that its development has had a profound effect on radiology and medical imaging. Whether hospital-based or private-practice based, informatics stakeholders agree that no matter what shape it takes in the future, this transformative tool will continue to evolve in ways that can only be imagined today.
PACS has, overall, been a “resounding success,” believes David S. Channin, M.D., founder and president of Insightful Medical Informatics, Inc. “The biggest achievement of the PACS era has been the remarkable conversion from film to digital environments, and the adoption of digital environment PACS- and workspace-based interpretation by radiologists, with significant improvements in efficiency and workflow,”
Channin says. He added that this has been accompanied by learned behaviors that have proven to be extremely valuable to radiologists, such as viewing a stack of CT images in cine mode, “something you couldn’t imagine in the 1970s.”
Channin also said that another major success of PACS is simply the fact that it has helped secure patient images. “From day one PACS solved the problem of ‘film, film, who’s got the film?’” he said, pointing out that PACS made images universally available within institutions even before the advent of electronic medical records and enterprise viewers.
And all of this has led to another truth about the impact PACS has had on radiology, Channin says, which is that it has become the leader in healthcare’s transformation into the age of the electronic health record (EHR). “We had PACS and workstations and digital printers that could print images very early on—long before everything else in medicine was electronic,” he notes. “One of the great successes of PACS was that you could digitize a complete environment in medicine, and that has been taken in a lot of different directions.”
As far as disappointments, there seems to be a general agreement that the biggest has to do with standards, or the lack thereof. “While everyone says they follow the DICOM (Digital Imaging and Communications in Medicine) standard, not everything is standardized,” notes Mike Quinn, chief technology and security officer at Jefferson Radiology, East Hartford, Conn., a radiology group with 10 private offices in central Connecticut, and which is affiliated with seven hospitals in the state. “It’s more proprietary. We’ve done a lot of migration, and—depending on the vendor—the migration can be very difficult to complete. And that impacts our ability to share information.”
Medical imaging has a “great standard,” in DICOM, Channin says. “We wouldn’t have been able to do anything in the past 30 years without it.” That said, Channin believes there are a number of ways in which the slow adoption of standards has been problematic when it comes to PACS.
Several stand out, the first of which has to do with interoperability. “We have a very good standard in DICOM, and we have very good HIE [health information exchange] standards for cross enterprise interoperability,” Channin says. In addition, he points out that the Radiological Association of North America’s Image Share program has demonstrated that image sharing can work across vendors and across institutions.
“But the commercial image sharing companies don’t federate,” he laments. “There’s no interoperability [between these commercial vendors], and that’s why we don’t have direct institution-to-institution image sharing. That’s a disappointment, and the only reason it’s not done is that vendors think there’s a competitive advantage in not doing it.”
Channin also laments the lack of advanced communication tools, specifically between various workstations and information systems within institutions. For example, if a surgeon using an enterprise viewer needs to ask a question of a radiologist who is using a radiology viewer, there is no easy way to do it, either through audio or video.
“It’s just really difficult to communicate,” he says. “Everyone is plugged into his or her particular applications. If you are a surgeon you might be plugged into your [EHR] module, and if you are a scheduler you are plugged into an [EHR] scheduling system, and if you are a radiologist you are plugged into a PACS.
“You can’t easily communicate person-to-person between those systems,” he continued. “But we can easily do that simply using our phones.”
The problem is that there has been no real business case for vendors to deal with these communication issues, he says, suggesting that that this could change with the advent of alternative payment models and bundled care. “There will be a [business case] when we are in these alternative payment modules and radiology groups are being paid for the care being given by 10 different providers.”
“There is a case to be made that before that surgeon orders a study, he should easily be able to communicate with a radiologist,” Channin continues. “And vice-versa.”
One issue ARA’s Thomas found troublesome was the that his PACS system “has been so locked into the Microsoft ecosystem. “It was very dependent on Microsoft technologies in order to function properly,” he says. “So we couldn’t do things like try to run it on a mobile platform or run it on a non-Microsoft system, or a non-Microsoft browser. That was a huge challenge for us.”
Like Channin, Thomas points to communication issues with PACS. “I think the lack of communication between disparate PACS—having to go to a third-party to get that accomplished—was shortsighted.”
He also notes that developments such as the evolution of the latest HTML standard, HTML5, are contributing “to a much more open system.” As a result, a PACS will be able to run on an iPhone or iPad or a Safari or Google Chrome browser.
“I think the classic PACS vendors are seeing the writing on the wall,” Thomas syas. “[They are] starting to move toward much more open systems.”
AN ANALYTICS COMPONENT
One of the more interesting trends, Thomas reports, has been the development of software that does quantitative analytics based on imaging data. “I think you’ll be seeing more of this,” he predicts.
For example, Thomas refers to the recent announcement about the creation of the Watson Health medical imaging collaborative, formed with 15 leading health systems, academic medical centers, ambulatory radiology providers and imaging technology companies.
Members of the collaborative plan to combine unstructured imaging data with a broad variety of data from other sources, and use Watson to extract insight to help improve how doctors diagnose, treat and monitor patients.
“There’s a lot of data there (in PACS),” Quinn agrees, adding that his group extracts data from its PACS, puts it into a data warehouse and uses it to perform operational functions, like running turnaround time reports. “This is important, especially with some of the hospital contracts we have,” he notes
Quinn also points out that his practice can utilize information extracted from its PACS to track radiologist RVUs. That data also is being leveraged for the benefit of population health management.
Channin believes that the Annotation and Image Markup (AIM) model and toolkit should be incorporated into PACS. The AIM project, funded by the National Cancer Institute, is the first project that proposes and creates a standard means of adding information or knowledge to an image in a clinical environment.
“This allows me to click on and measure something in a structured way with a controlled vocabulary,” Channin says. “Right now that information is captured in a very unfriendly way and isn’t structured or standardized. [AIM] needs to be adopted because that is one of the primary outputs of the radiologists—structured image annotation.”
Another area that Channin has been focusing on is deep learning, or artificial intelligence (AI), and how it will integrate into PACS, EMRs and other information systems. “Whatever your role is in a health system, you are going to sit in front of an information system that does your task with you,” he begins. “In the case of a radiologist, it’s a PACS workstation.
“Now, you’re going to have these decision-support modules or cognitive assistance, or whatever you want to call it, and how people interact with it will be very important. As they get more advanced, we are going to need standards for higher levels of interaction.”
He points out that this would be unlike a program like mammography CAD. That generates a set of DICOM images, annotated images and DICOM SR objects that are embedded and displayed in reports.
“When AI and cognitive assistances get more advanced than that—which is already pretty advanced—it’s going to be more interactive,” he predicts. “The question is how that is going to be integrated into the user interface of the clinician.”
A related issue, Channin says, is how these deep learning systems will change the nature of how radiologists think. “It will change behavior, and I don’t think we have a good understanding of how it’s going to change the nature of the daily interactions of the radiologist sitting at a workstation,” he notes.
BEHIND THE CURTAIN
Overall, our sources see the next iteration of PACS as more of an ecosystem than an actual system. “I think we’ll still be using the term PACS, but we’ll be talking about a ‘deconstructed’ PACS,” Quinn says, referring to an approach in which the core components of a PACS are broken down into “best of breed” solutions.
“You are going to see folks move to an image archive that has all the images from all ‘ologies’ that are connected and under a patient’s name, improving the ability to have all of that patient’s information together,” Quinn believes. “What you have for a worklist will be off of an EMR or an intelligent worklist and, finally, a viewer that will work across multiple modalities.”
Quinn also suggests that going forward, there will be more movement towards cloud-based PACS as well. This is an idea that others shared.
“PACS won’t exist like it exists today,” Thomas asserts. He again emphasizes that the VNA will become the “keystone” to any PACS solution.
“We’ll see a lot more vendors release independent viewers that will work with any particular PAC system,”he adds. PACS will be approached as a best-of-breed system.
“It won’t be ‘here’s your box, here’s your viewer, and here’s some sort of image management, storage, and worklist solution, all worked into one,” he says. “I think you’ll see that all break apart and become a lot more flexible for providers to use.”
As a radiologist and an informaticist, Channin believes that radiologists will eventually think about PACS basically as “the workstation app” and some sort of workflow director. “All of the backend stuff will be magic,” he suggests.
An analogy might be going into iTunes and either developing a playlist manually or having iTunes Genius suggest it, he says. “Either way, you hit play and you do your job of listening to the music.”
Likewise, Channin predicts, radiologists will interact with a PACS similarly. “There will be some kind of genius app that will see that Dave Channin is logged in and knows that he reads X-ray, CT and ultrasound, but doesn’t read MRI,”he says. “And it will start spitting out studies and capture those annotations and structured reports.”
Everything else—all of those intermediate workflow steps—the radiologist won’t have to worry about, Channin adds. “It will just be an app, and it will feed you information, present you with images, and relevant clinical context will be available to you—instantly.”