Digital imaging, telepathology: what’s the holdup?
The inter-linked fields of digital imaging and telepathology
continue to march forward, but more progress is needed before telepathology
will be used to perform primary diagnoses in real time, said Michael J. Becich,
MD, PhD, vice chairman of pathology informatics at the University of Pittsburgh
Medical Center and Shadyside/Hillman Cancer Center, during a lecture in April
at the Executive War College, presented by The Dark Report.
An equity owner and chairman of the scientific advisory board for InterScope Technologies, a manufacturer of high-speed, high-resolution whole-slide imaging robots, Dr. Becich began his April lecture with a disclaimer. "I’m not going to talk about the competitors’ instruments today," he said. "I’m going to give you a biased view. But I won’t make this a sales talk."
Digital imaging systems in their current state do provide advantages for laboratories, Dr. Becich says. For one thing, replacing nondigital gross-room cameras used for medicolegal cases and teaching has been a "grand slam in every laboratory that has put it in place," he says. "It’s far, far faster and more efficient than film. I’m sure we all recall the days of getting 12-packs of Kodachromes delivered in our mailbox in a batch, and them sitting in the corner gathering dust."
Unlike those Kodachromes, digital images do not have bracketing. That means pathologists can look at them on their monitors and capture them if they look good. "You get your feedback immediately," he says, and they mean "no film printing cost, no transport, no chemicals, and significant access abilities. For medical centers like our own, that do dozens of teaching conferences for surgeons, these are a requirement."
But digital imaging systems in their current state need not be as "ubiquitously installed" as the Department of Pathology at UPMC has done for the purpose of storing microscopic images in the 18-hospital system. "It was an unwise investment to go and buy all these cameras on sticks and make sure that pathologists in their offices or in a sign-out room have all these digital capture stations," he says.
"They’re costly to maintain, the pathologists don’t know how to use them, they’re not well integrated into the sign-out rooms, and although they are useful for teaching and conference support, there’s no reason to have one on every microscope, in every conference room," Dr. Becich says. "Having a centralized imaging facility where you can do the conference support is probably my recommendation today."
The chief problem with digital imaging systems in their current state is that they do not yet integrate well with laboratory information systems, Dr. Becich says. "What you end up with is people wasting their time annotating and storing and recording those images, off to the side of the work. That’s not billable," he says. "Unless it’s in the workflow and it’s recordable, with the LIS managing that information, I just don’t advocate that." He added jokingly: "They’re great paperweights."
The systems are useful for educational support, especially if they allow for multiple viewing modes. That way, "you can move them around the desktop like Powerpoint slides, gross and micro, and any other images you could import easily into the system," he says. "For conference support for residents, we have a laptop and a projector, they connect to the network, they never have to make film, they organize their slides generally 15 minutes before the talk, and can capture new images onto the system."
The systems have been less useful for placing images into pathology reports, though they do have that capability, Dr. Becich says. "It has failed in providing us a framework for taking images out for network-based or Internet-based reporting schemes. And that’s largely been vendor failure, not development failure. We have developed our own systems to do this. We can’t seem to get vendors to integrate those products and provide that level of service."
University of Pittsburgh Medical Center also has not found a vendor that can provide the ability to search against images. "You can search using the text in the pathology report and flag that there’s an image. We have not been able to get the vendor beyond that," he says. Being able to search the content of the images has been an active research area for his group. "But we don’t have a scalable commercial-grade system available for us right now—the idea being that you could put a query against the images, and say, ’Show me all the images that are chemical stains that are from men and that have high-grade nuclear features.’ And then query back those images."
Vendors have begun integrating the option of putting digital images into reports, Dr. Becich says, but not usually as part of LIS functionality. "There are several products hitting the marketplace that are promising Web-based physician facing and pathology facing reports, but progress looks real slow to me in terms of getting how to use this in real time," he says. "They’re put in by integrators or re-sold through the existing vendors through Web portals," which isn’t as satisfying, he says, as being able to turn on a switch and say, "Create a Web-based report, and make it available to my clients."
Dr. Becich would like to be able to better integrate anatomic pathology, clinical pathology, and molecular diagnostic data, which boutique vendors have begun to do for the outpatient market. But pathologists themselves need to work harder on fostering that kind of integration, he says.
"There’s that six-foot-thick wall between our anatomic and clinical pathology
laboratories, and very little communication," Dr. Becich says. "We’re the biggest
part of our own evil, in getting to something like this that really integrates
our own data." Pathologists are comfortable with what he calls the "diarrhea
of reports"—the molecular diagnostics report, the hematopathology report,
the heme report, the pathology report, the bone marrow report, the blood smear
report—and never put it all together. "And that’s really starting to nip
at our heels," he says, "if we don’t figure out a way to do this kind of reporting,
which puts everything in the context of the disease for the patient."
The power of that sort of integration can be seen in his department’s initiative to post on the Web cases that residents present at conferences complete with digital images. Since 1994, the department has posted about 400 cases, and the site gets about 1.2 million hits per month from 85,000 unique users.
"Dozens, probably hundreds of schools are using it," he says. "All sorts of para-professional schools, medical schools, are using this. We get routine requests for folks to download these on CDs and create shadow sites for them. They’re designed by pathologists for pathologists to teach pathology."
This framework shows the promise of what labs could be doing with what Dr. Becich terms "the majority product of your pathology laboratory: your report." But the lack of electronic medical record integration presents a threat, he says. "Through the use of e-mail, through the use of the Internet, through the use of other tools, we need to get back in the queue and deliver a customized report to our clinicians. I think these imaging initiatives are key, at least in anatomic pathology, for that to occur."
Well-developed digital imaging is critical,
too, for telepathology, which also suffers from a lack of integration with the LIS, according to Dr. Becich. "The majority of systems that are out there today are static, and Web-based, meaning stick on a camera, snap a picture, send pictures upline, either FTP them or send them in a wrapper that somebody opens up and looks at asynchronously," he says. "Meaning you send it at 2 o’clock, it gets there at 4, it may sit until 6. Somebody then gets a flag that it’s there, they look at it, and they shoot you back an answer."
Partly as a result of this, the most successful telepathology operations handle hundreds of cases per year, Dr. Becich says. "And second-opinion consults in the United States are estimated at 1.5 million per month. So a successful telepathology system with hundreds of cases ain’t so successful." At the University of Pittsburgh, he and colleagues have been doing telepathology for five years, and their numbers are "barely approaching 1,000 now," he admits.
Among other problems is the fact that such static systems require a small sub-sampling
of the diagnostic specimen. "Dynamic telepathology systems where you real-time
view a slide over a network with a dynamic connection have been too costly and
cumbersome to use—and again, aren’t integrated through the LIS," Dr. Becich
says. The University of Pittsburgh attempted to set up a Web-based static system
starting in 1997, he reports, but "we have not had a lot of success with either
getting clients or building a book of business." That system was recently shut
To get past the problem of sub-sampling, Dr. Becich’s department attempted to build a dynamic telepathology system, in which a slide is loaded onto a robotic microscope with high-speed network connections. "The problem with that was, again, the network connections were god-awful expensive, as were the robotic microscopes, a quarter million a pop."
In conjunction with the University of Pittsburgh, InterScope Technologies has developed a hybrid system that has had "moderate success," Dr. Becich says. "We capture images, and we have over just telephone-level connections the ability to transmit the entire case as an integrated case with ties to the LIS, to the remote site that does the diagnostics." They can report in the system, chat if they bring the other person online in kind of a chat-box tool, and then report the case. And they can hold case conferences, a feature designed for a UPMC-owned hospital in Palermo, Italy, that did not have transplant pathology expertise.
Once a case is put into that system, it sets off a beeper so the on-call pathologist knows there is a case ready for viewing, which he or she can find queued up and ready to view on the screen. "It’s kind of real-time in that when an event happens, somebody’s triggered to go then and look at a case," Dr. Becich says. "The transmit time between Italy and Pittsburgh, for an average case, which has 12 images in it, is about 11 minutes."
In place for more than four years, the system, which includes case demographics and history as well as the ability to report and bill, is approaching 1,000 cases. "It allows ... the ability to chat real-time, if you’re viewing the case and you want to ask the question of somebody on the outside," he says. "It records all the conversations that you have in the chat as part of the record, and then seals down the final diagnosis and walks it down like it was a case."
The system’s limitation is that it still transmits only a very small sub-sample,
whereas pathologists would much rather be looking at the whole slide. "Generally
the person who is providing you the images as a consultant doesn’t know what’s
important, or they wouldn’t be consulting with you at all," Dr. Becich says,
"though they may have a good feel for what’s important." But in a pathology
case, if there are 15 slides and they’re sending 10 snapshots, "there’s a great
degree of ’disease,’" he jokes, "about the disease in terms of interpretation."
Three companies are in various stages of developing whole-slide telepathology systems, Dr. Becich said, citing InterScope, Aperio, and Trestle. Such systems enable the pathologist to capture the entire slide at high resolution and display it at multiple resolutions, with the goal of imaging all slides in a multi-slide case, not just one or two.
"Clearly the focus is to integrate this with the LIS in terms of my own principles for how this has to work," he says. "And it needs an intuitive pathologist-like interaction so you can easily navigate and use the images." It needs to run on standard networks and browsers, it needs to have an imaging format that allows the pathologist to take the image and do whatever else he or she wants with it, such as integrate it into the electronic medical record. "And it has to be affordable," he adds.
InterScope’s system uses optical components and tissue-finding algorithms that
allow the user, in high-speed, to focus through the entire glass slide and create
a well-focused image. "This is done through continuous stage movement," Dr.
Becich says, "which is different from our competitors, and also through custom
imaging software that we’ve developed to then view the slide image along with
the LIS information—all within the framework of the pathology workflow."
The system uses an ultra-rapid digitizer called "Xcellerator" that applies a "montage image capture technique," Dr. Becich says. "The robot identifies the slide from a bar code as belonging to a patient and being a particular block. It then identifies on the slide where the tissue is, focuses in terms of flat fields, multiple flat fields across the slide, and then essentially goes field after field until all the tissue has been imaged." It then montages all that back together and provides it as a whole image. "It’s not a scanning. It’s actual snapshots. It’s individual pictures," he says.
Such a system allows pathologists to be able to take all the slides from the laboratory, image them, and provide them through a storage area network back to the department. Multiple pathologists can view the slide at the same time, experts receive specialty cases like skin and GI path, and general cases are distributed to trainees or other nonspecialists.
The system will "help us get over the barrier of sharing slides across distances," Dr. Becich says. "So if you had a pathologist in 18 laboratories like we have, and we want to centrally process and then distribute, it’d be nice to give the right kind of work to the right kind of people rather than just one lab by one lab, 18 labs, sending all of the complex mixtures, and having couriers run around. It would give access to experts irrespective of the patient location. It would obviously allow you to prioritize caseload and reallocate the cases I talked about. It would provide immediate access to archived images."
For now, the approved use for most telepathology devices is for "second opinions, consultative diagnostics," Dr. Becich notes. The pathologist using the telepathology system is looking at it for a second opinion on an already rendered opinion. But the prospect of its use in primary diagnosis is where he thinks it "gets a lot more interesting," from the standpoint, he says, of how it can affect pathology practice.
Ed Finkel is a writer in Evanston, Ill. InterScope’s Web site is at www.interscopetech.com.