Internet Teleradiology: The Other PACS

Teleradiology is the most widely deployed application of telemedicine. In our last annual survey of the field, reported last October (Vol. 5, no. 5), we documented that there were about 22,500 interpretations being done each month in the U.S, or about 270,000/year. About 90% of these were diagnostic, and about 10% interim-reads. Another 4,500/month were being done in Japan, and 500/month in Norway. These figures were over double those of the year prior, and dwarf the figures for interactive consultations (see story, this issue).

In use since the late 1950’s, teleradiology is the "granddaddy" tele-imaging application. Numerous definitive studies (see Twenty Selected Teleradiology References, Vol. 4, no. 2) have shown that transmitted radiographic images (with the possible exception of mammograms; see Telemammography Feasibility, Vol. 4, no. 2) can be displayed on a remote monitor and interpreted with diagnostic reliability.

In the past year teleradiology, which refers to the transmission of images between enterprises, has begun to integrate with PACS (Picture Archive and Communication Systems), which refers to image acquisition, management, and transmission within a single enterprise. The lines have been blurring as health care institutions link together their far-flung affiliate hospitals and clinics.

More recently, radiology managers and clinicians have started to explore the use of the Internet for transmitting images. Because of the universal protocols (TCP/IP) used, and the very low cost of bandwidth, Internet technology promises profound, far-reaching changes in the way medical images are used and managed. These changes will accelerate as Internet 2 – a more secure, reliable, and expensive grade of Internet service – becomes available.

This story examines the ways nine key teleradiology vendors are integrating the Internet into their product lines. It also features an interview with Dr. Jim Logan about the pitfalls of putting together a teleradiology system that really works.

Decision Making for Teleradiology

An Interview with Dr. Jim Logan

[TT interviewer: Ace Allen, M.D.]

Jim Logan, M.D. is President of Logan & Associates, Inc., one of the original telemedicine consulting firms. Their client list includes numerous public/private, rural/urban hospital-based health care delivery systems. In addition, they have provided services to medical organizations in Russia, Japan, Australia, Iceland, Scotland, France, Germany, Canada, Guam, South Korea, and the Peoples Republic of China. Dr. Logan completed a NASA-sponsored residency in Aerospace Medicine and is board-certified in the specialty. Since last August, he has served as the Telemedicine Clinical Director for the U.S. military's Pacific Regional Program Office (PRPO) based at Tripler Army Medical Center in Honolulu, HI. PRPO is a consolidation of the AKAMAI Telemedicine Project and PACMEDNET. One of his interests is finding low-cost ways of providing diagnostic teleradiology services throughout the Pacific Basin, a sparsely populated area that covers roughly a third of the planet's surface.

TT: Jim, what problem is the Department of Defense (DoD) trying to solve by using teleradiology?

JL: The DoD has military and non-military medical beneficiaries in the Pacific Basin and beyond that include active duty personnel, their dependents, and limited civilian populations such as contractors, etc. The problem is really quite simple – timely access to remote medical expertise. Many DoD facilities (including ships at sea) have only basic healthcare capabilities and are hundreds if not thousands of miles from the nearest specialists, including radiologists. The turnaround time for getting back a professional interpretation for a plain film study can be as long as six weeks. The logistics are daunting and the entire process can be cumbersome and very expensive. Today, believe it or not, films are actually flown halfway around the world from Diego Garcia, a base in the middle of the Indian Ocean, to the naval hospital in San Diego via the medevac system.

TT: Other than time and geography, are there other issues teleradiology can address?

JL: Absolutely. Unfortunately, the military has been very slow to realize it is actually one giant worldwide medical enterprise. The different services and various commands have been throwing technology at the problem and expecting it to stick. At NASA, I was Chief of Medical Operations. I learned over and over technology isn’t the answer, it’s just a tool. From a tactical perspective, telemedicine can positively impact several operational models such as mission readiness, transportation (medevac), remote electronic triage, and medical consolidation.

For example, in a true enterprise environment, teleradiology could facilitate real time "workload leveling" to even out situations in which there are too many images and too few local radiologists. When we performed a Telemedicine Needs Assessment at the U.S. Naval Hospital in Okinawa, I found the radiologists were reading images taken 18 days previously. Right before I interviewed one of the radiologists, he had found a fracture of a little bone in the hand for one of his fellow-physician’s daughters in an image that was almost three weeks old. We calculated there were approximately 8,000 routine studies a year that should be offloaded to other DoD radiologists who aren't as busy. These radiologists could be anywhere.

There are other problems as well. The military’s medical information system, known as CHCS (Composite Health Care System), is just awful in my opinion. It doesn’t support images, graphics, audio, or video, just text. It’s a classic multimillion-dollar legacy data repository system that has little, if any, interoperability capabilities between various DoD sites. In other words, CHCS at Tripler can’t interface with the CHCS system in Guam without an expensive workaround that is unique to those two systems. Even the workaround has to be customized depending on which two systems you are trying to interface. In Okinawa, once the radiology report is dictated, there is up to a two week delay in transcription. Then, reports are issued within CHCS. So, if you are a physician at the Medical Branch Clinic in Sasebo, Japan and your films are being read by radiologists at the U.S. Naval Hospital in Yokosuka (800 miles away), you have to telnet into the CHCS system at Yokosuka and go fetch the report. All this expensive technology and the poor doc has to go fish for his data. This is a terrible misuse of physician time. It makes an operations guy like me shudder. It’s definitely not technology serving the user. At the very least, reports should be "pushed" back to the referring physician via email or something. I’m told there is a 2-year backlog for change requests for CHCS, such as activating an automatic fax of the radiologist’s report within CHCS. This is just one example process engineering problems that plague telemedicine care delivery.

TT: How important are "process engineering" issues?

JL: Critical. As is always the case in telemedicine, there are many more operational and organizational barriers than technical ones. Too often the clinician isn't an integral part of the decision-making process. Telemedicine has always been episodic. To make it mainstream, you must incrementally shift the medical "center of gravity" from "Usual Care," which consists of face-to-face interactions between patients and specialists along with significant patient transport burdens, to "Distributed Care Environments" characterized by care at a distance, remote electronic triage, and "virtual" departments functioning seamlessly across a medical enterprise. In short, the goal is to "Move Bytes, Not Bodies" whenever possible. To make this shift, you really have to think about workflow and workload issues. This requires very careful assessments of the real value of physician time and, in the case of teleradiology, technician time as well. Although this is happening to some extent in the PACS world, it simply isn't happening in the teleradiology world or the telemedicine world in general.

TT: Why not?

JL: It takes a lot of work to change the paradigm. Another thing I learned at NASA is that people manage at their level of understanding. In telemedicine, people only understand the technology so that’s the level at which telemedicine is managed. This is why telemedicine is still basically an Information Systems initiative even though everyone agrees it should be clinically driven. You have to move beyond the work of selecting the technology, to the much harder work of figuring out how to make it serve the operational environment rather than make the operation serve the technology. If you don't do that, the technology becomes just one more barrier the physician has to hurdle to effect care. Most telemedicine projects just deploy technology, then wonder why it’s not working. Utilization is still the "dirty little secret" in telemedicine. Frankly, it is much easier for telemedicine managers to be "boys with toys" than it is to be strategic process engineers. Look at the military. They’ve spent millions on telemedicine hardware but have done little to re-engineer their environment to incentivize the rapid adoption of telemedicine. A military specialist can’t even get workload credit for a store and forward consult. In my experience, you can’t take a crowbar and wedge technology into a static rigid operational environment. It simply will not work. That approach is futile and conflicts with my "Telemedicine Prime Directive."

TT: What is "Logan's Telemedicine Prime Directive?"

JL: It is this: Changes in the technical environment require even greater changes in the operational environment. Any program that doesn't recognize this isn't going to fully leverage the benefit of technology. Teleradiology should be a tool to increase productivity and quality. The way you do that is by automating, or semi-automating, the workflow.

TT: Give us an example of what changes must be made in the standard radiology environment to make teleradiology work.

JL: Here's an example of workflow using the "old" x-ray viewing process. A technician goes to the "in basket" and the film library, assembles the required studies, brings them to the reading room, and slaps them up on a rollo [a sequential motor-driven viewbox that can "roll through" dozens of studies lined up four abreast - Ed.]. The radiologist cycles through the images, going forward and backward as necessary to compare images, makes a diagnosis, then dictates the interpretation. The interpretation is transcribed, printed out, reviewed by the radiologist who signs off, then it is affixed to the patient chart or mailed or FAXed to the referring physician. Later, the images are removed from the rollo and taken back to the film library. This is a pretty efficient system, so far as it goes. It maximizes the use of its most expensive unit - the physician. However, it is completely useless for reading off-site images in a distributed care environment.

TT: What should the workflow look like in an optimized teleradiology scenario?

JL: There are no simple, universal solutions. Each physician, administrator and organization will have to think out their own solutions. Here's a start, though. It should be simple and the work should be automated or semi-automated behind the scenes, like this:

1. A technician at the local site scans and digitizes the hard copy;

2) On-screen, the digitized images are labeled and the recipient's electronic address is attached;

3) When it is time to send the individual image, or batch of images, a single mouse click encrypts, compresses, and routs them to the remote specialist;

4) Pertinent patient history accompanies the image, in digital format;

5) The actual transmission is automated so that, except for contingency reads, it sends files when telecommunications links are cheapest - typically after hours;

6) On the receive end, the images are automatically assembled into patient folders, which could include previous studies;

7) They are then automatically queued up as the radiologist prefers: by case, date and time, diagnosis, etc. on the radiologists receive station;

8) Then and only then, the radiologist is alerted, by a screen message or beeper, that images have arrived and are ready for interpretation;

9) Images can be compared side-by-side, as they are now in PACS systems.

10) Upon finishing, the radiologist dictates the interpretation, perhaps into a voice-recognition system.

11) The dictated report can be checked for accuracy on the spot, then routed automatically via email or FAX to the referring physician.

11) Completed studies are routed to short-term storage, which rolls into long-term archiving if they aren't accessed within a specified time period.

13) Ideally, everything is in a nonproprietary format, so the radiologist could forward an image, perhaps annotated, via email to any provider on the planet if indicated.

Of these baker's-dozen steps, only the first one is unambiguously here now. The rest are in various stages of development. A few of these steps are now integrated into PACS delivery systems, but haven't quite made the transition to the low-bandwidth, low-cost teleradiology world.

TT: So what do we have now, in terms of teleradiology?

JL: We have systems that work, but are too awkward, require too many steps, and too slow to become mainstream avenues of health care delivery.

TT: What about the speed issue? I thought that new compression algorithms, such as wavelets, had made bandwidth requirements much lower - and thus, transmission speed much higher.

JL: Yes, that’s true. But the big problem is that many teleradiology systems require that the image be compressed at the server level, rather than at the acquisition (scan) station level, the so-called "client." That's fine when the server is in the same building, but it may not be. For example, at the Medical Branch Clinic in Sasebo, Japan, images would have to be sent uncompressed to the server at Yokosuka, 800 miles away. Our Telemedicine Needs Assessment revealed there is literally not enough time in the day to transmit the required number of images if limited to the bandwidth currently available. One solution we looked at would have required either an onsite server costing almost $30,000, or expensive broadband leased lines or satellite links to transmit the images. It became obvious we needed a system that could acquire and compress images on the client without a synchronous connection to a remote server.

TT: Have you found a system that comes close to what you're looking for?

JL: There are a lot of good vendors working on these problems. You have to be rigorous in pinning them down as to their product's capabilities. For example, can they really compress on the client? Also, the software is developing rapidly. I suspect that by the time readers see this, some of the 13 steps I outlined earlier will have been addressed.

TT: Final comments?

JL: The basic teleradiology technology is state-of-the-art. It works. It is dependable and accurate. It now needs to be truly integrated into the real operational environment of radiologists. The technology has to make their work faster and their lives simpler, not more complicated. It needs to do for them, in a distributed environment, what a rollo, technician, and other support staff do for them in the usual care environment. The larger telemedicine world in general needs to learn from the teleradiology experience. As long as the emphasis stays on the "tele" rather than the "medicine," there will always be a lot more work to do.