Healthcare Imaging Strategies Not Exactly a Snap

This article first appeared in the January/February 2013 issue of HealthLeaders magazine.

Talk about enterprise imaging, and you're likely to witness some pained expressions from healthcare CIOs.

In recent years, the healthcare information technology industry has been particularly innovative when it comes to creating images. Picture archiving and communications systems are in every radiology department, and mini-PACS have proliferated in many other specialty departments, everywhere from cardiology to pathology to dermatology. Meanwhile, tablet computers equipped with 5-megapixel cameras are in the hands of a large number of shutter-happy clinicians.

The vendor industry's response—enterprise imaging—largely amounts to a series of promises waiting for implementation at all but a few pioneering organizations. According to a July 2012 survey by KLAS, most healthcare providers are just in the early stages of forming an enterprise imaging strategy.

"This is both an exciting time and a little bit of a confusing time for enterprises. Vendors have recognized that there is a need for and an emerging marketplace for 'vendor-neutral' archives that will handle not just radiology DICOM images, but all kinds of documents and images from multiple disciplines, everything from photos and light photography to pathology slides and ophthalmology images," says David S. Mendelson, senior associate in clinical informatics, director of radiology information systems at Mount Sinai Medical Center, a 1,171-bed tertiary care teaching facility in New York City. He also serves as cochair of the board for Integrating the Healthcare Enterprise International.

"Enterprises, through their CIOs predominantly, have said that there is an economy of scale by trying to have one central archive ultimately that would handle all these things. Intelligent applications sitting behind that archive would enable the ability to distribute all those various types of images in an intelligent manner to the right people," Mendelson says.

"You need some authentication rules behind it. Not everybody is entitled to see everything, but if you have everything aggregated in one place, it may become less expensive to manage. It may become less expensive to build a business continuity or disaster recovery system behind it. Then what you need is a set of business rules for who can look at what and when."

Traditionally, PACS applications and PACS archives were closely integrated. In an enterprise imaging model, the archives—the storage and the information associated with the stored images—are often separate from the applications. At least, that's the ideal. Between reality and the goal lie a host of issues. Existing PACS archives have to be upgraded to vendor-neutral archives, able to store and serve up images in all their many file formats. A fast, powerful network architecture must be able to scale and yet provide assured levels of service, both in terms of availability and latency. Rule sets established at the archive must be enforced across an ever-growing toolbox of imaging applications.

"For those of us, particularly radiologists, who have existing archives, the cost of data migration into a vendor-neutral archive is a real factor that has to be taken into account," Mendelson says. "Not just the cost, but the process and the time it takes."

At Mount Sinai, the strategy to achieve all these aims is, of necessity, an incremental one. "Right now in many places, electronic medical records are eating up all the dollars for IT projects, which means vendor-neutral archive only gets a middling priority, and that may be very appropriate if you're functional right now," Mendelson says.

"Some places are in the process of replacing PACS systems," Mendelson says. "That might be the moment to begin to entertain moving toward a vendor-neutral archive instead of going to a traditional PACS model."

Mount Sinai is evaluating migrating to a vendor-neutral archive, which would also lay the groundwork to allow the organization to provide images to health information exchanges without increasing the vulnerability of its primary imaging applications, he says.

The vendor-neutral archive may serve double-duty as an upgraded business continuity and disaster recovery system, replacing a cumbersome magneto-optical disk backup system, he says. Planning, and lots of it, is the place to start with enterprise imaging.

"Unless you have a comprehensive enterprise strategic plan for enterprise images, you will waste a lot of money," says Paul Chang, MD, FSIIM, professor and vice chairman of radiology informatics and medical director of enterprise imaging at the University of Chicago School of Medicine. "You're going to consume a lot of cycles, because it is unsustainable having this menagerie of various mini-PACS without a comprehensive management plan. And unless you have optimized work flow, you won't leverage and achieve the efficiencies."

Chang says the often-heard goal of integrating DICOM and non-DICOM images into a single image archive is a solved problem at scores of medical centers. The real need, he says, is for mature work flow technology that allows those capturing the image to enter necessary metadata about the image at the time of acquisition.

One such application, built at the University of Texas MD Anderson Cancer Center, which conducted more than 10 million diagnostic imaging procedures in fiscal year 2011 at its 594-bed hospital, was demonstrated at the 2012 HealthData Initiative Forum's Health Datapalooza event in Washington, D.C. The application, ViSion, was described at the event by David J. Vining, MD, medical director of MD Anderson's Image Processing and Visualization Laboratory. "What we've tried to do, though, is capture the natural work flow of a radiologist, and that is simply sticking our finger on the film and saying, 'Where is it and what is it? Like, lung cancer? Colon polyps? Gall bladder? Gallstones?' And that's the essence of what we have done with this app."

Application development is not the typical hospital's strong suit. Chang cites two government agencies—the Air Force and the Veterans Administration—that have built metadata-at-image-capture applications. The rest of healthcare is still waiting for imaging system vendors to offer mature solutions of their own. It doesn't help, Chang says, that some vendors are still trying to get providers to buy bundled archive systems, even as those providers have invested heavily in their own storage solutions.

"One of the problems with non-DICOM images is it was too easy for us to just throw them into a PACS without really associating with metadata or orchestrated work flow to be useful," Chang says. "You've got all these pictures, but they're not as discoverable, they're not as useful to us as radiology images, because of the lack of association with metadata.

"You might, if you ever get around to it, tag [or] associate with some metadata post-hoc. That's very dangerous. We would never do that in radiology."

At the University of Chicago, Chang is building an iPad app that integrates work flow into the act of taking a picture. A nurse opening the app will be able to obtain a patient list, select the patient being photographed, and use drop-down menus to indicate the target, such as a skin lesion. Then the application will give the nurse pointers for how to take the picture.

But a total solution takes a lot more than building one iPad app. The University of Chicago first built a sophisticated service-oriented architecture infrastructure, and could rely upon Chang's earlier experience at the University of Pittsburgh, where he helped build the Air Force's system that was under contract at the time.

All of which indicates that solving the enterprise imaging puzzle requires expertise, years of time, and money. It's more appropriate to think of enterprise imaging's potential to create value than it is to think of it simply as a potential return on investment, says Eric Yablonka, vice president and chief information officer at University of Chicago Medicine.

"It's hard to quantify the ROI, but we can tell you that it's created value," Yablonka says. "It's significantly improved our turnaround times and our service to patients, and it's improved our teaching program. It's allowed us to more quickly and with better accuracy report the quality of our imaging service, so it's not just return on investment. I would say the same thing to anybody else who wants a strict ROI interpretation of any IT investment. It's about creating value."

Enterprise imaging "is an opportunity both to leverage cost and infrastructure, but also to deliver integration and work flow to the EMR, and I think that's critically important," Yablonka says. "It is really critically important to have high-quality, very productive imaging groups, and whatever we can do to improve their work flow using technology, we should."

As information flows increasingly from provider to provider via health information exchanges or accountable care organizations, enterprise image management will be crucial, adds Yablonka. "You don't necessarily want patients to be imaged over and over as they move through different phases of care, but you then have to have an image sharing platform, almost like an image HIE kind of capability, and traditionally images have not been core to HIE plays. So I think that's a huge opportunity going forward on the imaging side to both save money in the industry and reduce exposure to patients of excess image radiation and to improve the quality of care. That's going to be very big."

So far, the Centers for Medicare & Medicaid Services has not made access to images through an EMR a requirement for providers. But meaningful use Stage 2, the rules for which were finalized in October 2012 and which take effect in October 2014, includes for the first time as a menu option the requirement that EMRs be able to access more than 10% of all test results of which is one or more images ordered by an authorized provider of the eligible hospital for patients admitted to inpatient or emergency department. And CMS has a track record of incorporating menu items in one stage of meaningful use into the core requirements of the next stage.

"Meaningful use says it doesn't matter if this patient had a chest x-ray from another hospital or another PACS," Chang says. "You need to be able to display it on this other EMR. That's a nontrivial issue. This is a huge issue. So the real challenge of enterprise images is not the enterprise as we know it today. The real challenge of enterprise imaging will be when the definition of the enterprise goes beyond our firewall."

At 550 beds, Chang says Chicago is not a big system. "We don't have that problem. But trust me, a lot of folks call me all the time asking for my advice on how to address that nut, because it is a really difficult problem when you get beyond a simple definition of the enterprise."

Realizing that vendors alone will be challenged to solve this problem, the National Institutes of Health recently renewed funding for RSNA Image Share, a project headed by the Radiological Society of North America. Mount Sinai's Mendelson is the principal investigator.

Initially aimed at sharing of radiology images, RSNA Image Share builds on the IHE XDS-I profile, a solution for publishing, finding, and retrieving imaging documents across a group of affiliated enterprises. The preliminary  focus is for patients to control the routing and exchange of their imaging exams through image-enabled personal health records. In its first pilot stage, the project already empowers more than 2,000 enrolled patients to direct how such images will be shared. "That cuts out an entire layer of security and confidentiality concerns, because in a health information exchange, the patient has to be consented and give permission," Mendelson says. "There's a whole layer of bureaucracy around that, and there's good and bad in that, but this other methodology would let patients build their own imaging records in a personal health record that's been image-enabled. The patient then directly controls image distribution."

Looking even further ahead, there may come a day when providers and patients can browse relevant images on a tablet in a manner vastly easier than today's medical record. MD Anderson's ViSion project sports a user interface inspired by the Operation silly skill game, where images are laid on the outline of the human body, and physicians and patients can flip back and forth through images to see a timeline of disease and treatment.

"We are now actually integrating treatment icons, so that we can show when certain treatments like surgery or radiation therapy have been effective and might have affected the course of disease in terms of the images," said MD Anderson's Vining at the Datapalooza event.

Best of all, the images are brought into ViSion by that lowliest of PC-based imaging technology: the screen capture, "whether it's a PACS system or a 3-D imaging workstation, and we extract the image off of this screen capture and upload it to a cloud server," Vining said. "As I talk about this image with my microphone, I capture this image and my voice, I upload it, and from that voice we extract metadata to tag that image, but because we use screen captures, we integrate with no one vendor but we interface with all of them."