Teleradiology -- A Personal View

Introduction

Teleradiology is already a fact of life in many hospitals and, to many, heralds the age of telemedicine. Using teleradiology, digital images can be sent throughout the hospital and to radiologists at home for on-call reading.

The perceived benefits are many. From the hospital point of view, there are financial savings if "moon-lighting" or part-time positions can be eliminated. For radiologists who live far away, teleradiology can mean giving diagnoses from home without have to drive to the hospital. For the referring clinician, it can mean more rapid assistance and better patient care. For hospitals without a full time radiologist, it can mean coverage where none existed. It can mean consultation with experts and training at a distance. Few technologies, in my experience, have been embraced so thoroughly and with such consensus as teleradiology.

As a staff radiologist with a background and interest in computers, I was asked to be involved with our hospital's selection of a teleradiology system. Now that our selection has been made, I interpret emergency cases using our teleradiology system while on call.

Because the impact of teleradiology is potentially so pervasive, it is important for physicians in general to be involved so that the ultimate selection of technology will meet the needs of the hospital as a whole.

Basic Information

In teleradiology, the center of the operation is the hospital server which stores the clinical studies that are the source of images that radiologists at home will review and what clinicians in the hospital will see.

If clinical studies are already digital, as in the case of CT (x-ray computed tomography) and MRI (magnetic resonance imaging) data from the imaging consoles can be fed directly to the server and thus be available for radiologist interpretation. Images can be windowed and leveled just as at the console.

If hard copy film, such as a chest x-ray, is to be interpreted, it must first be made digital by scanning it into the computer system. This is similar to the flatbed scanner which many of us use to scan documents into our computers but differs in that it uses transmitted rather than reflected light and has much better resolution. The digital image thus created can have its brightness and contrast adjusted but cannot be truly windowed and leveled.

No Problems Anticipated

From a technological point of view, the main processes of teleradiology are almost trivial. Most of us with computer know- how have been uploading and downloading images and programs from BBS's (Bulletin Board Services), using them, storing them and sending them to each other long before the Internet became part of our common vocabulary.

From an institutional/organizational view, there should be no problem in introducing a technology everyone wants.

So, What Is The Problem?

The basic problem is that even a change for the better is change. There is more work involved for the technologists who generate the images and more work for the radiologists who have to get used to a new way of looking at images. Not everyone is comfortable around computers and skills must be learned, often on personal time.

Then there are the expectations of the referring physicians and the hospital regarding the impact of teleradiology in improving care and saving money.

Teleradiology is relatively new for many physicians. Vendors are new, the instrumentation is new, and the language is new. Little wonder that there is confusion and uncertainty.

Selecting a System

If on-call radiology coverage is the major goal, it is not immediately obvious that every hospital needs its own teleradiology system. There are alternatives ranging from in-house coverage to using a teleradiology service provided by others.

Even if owning a teleradiology system makes the most sense economically and in terms of patient care, equipment selection will depend on the goals and finances of the radiology department and the hospital. Most likely, teleradiology will be but one of several projects hospitals will be considering to try to remain competitive.

It seems obvious that those who will be using teleradiology and those with computer experience should be somehow integrated into the decision making process. Astoundingly enough, purchasing decisions are often made by those who know the least about actually using the equipment and, indeed, will never use it themselves once purchased.

Unless a hospital is going to develop its own teleradiology system, it will have to make its choice from commercially available units. Unfortunately, there is no "Consumers Report" for the instruments we buy.

In the absence of objective reviews, one can attend a meeting like that of the RSNA (Radiological Society of North America) in which many teleradiology vendors can be expected to present. Ultimately, there is no substitute for "hands on" testing and finding out, in great detail, the experience of others. I would suggest that contact must be with both the people who perform teleradiology and those who have to live with the results.

Finally, any equipment selected for purchase must be shown to meet performance criteria on site before it can be accepted.

Integrating Teleradiology Into The Hospital Environment

The current buzzword in Radiology is DICOM (Digital Imaging and Communications in Medicine) networking compatibility which allows equipment from different manufacturers to "speak" with each other. Actually, the DICOM standard is complex and is still being developed. Because of this, and it is possible for equipment to meet the DICOM standard in certain respects and not in others. Therefore, vendors must clarify what they mean when they say their equipment is "DICOM compatible."

Radiology is not the only service dealing with image interpretation and storage. Cardiology and Pathology are moving toward computerization and face similar issues. It is not clear if they will adopt the DICOM or a DICOM compatible standard. Thus if pathologists and radiologists would like to review each others images and reports, interdepartmental connectivity is required.

Proprietary vs. Open Solutions

If a hospital chooses a vendor with proprietary hardware and software, it becomes dependent on a specific vendor for even the most minimal modification. Proprietary solutions tend to be rigid and are particularly vulnerable if the supplier goes out of business or is taken over by a larger company.

A more flexible approach, for example, might be to store patient images on the Internet or on a local intranet in a standard format, possibly with encryption for safety. Then any physician with a computer, a browser, and viewing software -- and the appropriate password and decryption key -- could access and view studies of his or her own patients.

As teleradiology matures, "shakeouts" will almost inevitably occur and those with the most open and flexible systems will do best.

Equipment Issues

It is said that one can never have too much ram (random access memory) or too much hard drive space. This is true. In hospital, a major issue is long term (archival) storage. I cannot overemphasize how important having enough storage space is. On home workstations, storage is less of a problem because studies can always be retrieved from the server.

Speed of image transmission between hospital and home is something that vendors dwell on, skipping freely in their discussions between 56k modems, ISDN lines, ADSL, and a few other acronyms Of course, transmission speed is important, but is only one factor.

Time for compressing and decompressing images is typically glossed over during sales presentations. Typically, images or image data will be compressed at the hospital server before being sent to the radiologist. Images will then be decompressed on home workstations prior to viewing. This can take significant time and computing power.

There are other issues. If the hospital based teleradiology unit cannot scan images, compress data and transmit images at the same time, it will take longer for it to send a case than a system that can perform several functions at once.

Similarly, if the home workstation grinds to a crawl as new images are being sent and decompressed, the radiologist cannot smoothly review images and, again, the whole operation slows down. The ability of computer systems to perform two tasks seemingly at once, called preemptive multitasking, is an important purchase consideration.

Conversely, extended discussions about operating systems like Unix, Linux, BeOS, and NT are interesting but only in so far as they address the basic issues of speed, stability, and usability.

What Should The Software Be Like?

At hospital, software should be intuitively easy to use, essentially crash-proof, with lots of feedback to the operator as to what is happening. The user should be able to scan, compress, send, and monitor everything all at the same time. Not all feedback need be visible but it all should be easily accessible. If a telephone connection is dropped or transmission is incomplete, data should automatically be resent and the operator be notified.

Extensive logs need to be generated so that problems can be investigated and the efficiency of the system can be assessed over time. As far as possible, the system in the hospital should be able to monitor the computers at home.

Home workstations should be powerful enough to function quickly, without being hindered by simultaneously transmitted cases. Software must be stable. The radiologist should know when a case is being transmitted along with such details as the patient's name and the total number of images in the study.

The list of cases in the home workstations should update itself automatically without the radiologist needing to leave and then re-enter the image viewing program. Studies should indicate the patient's name, date of examination, time of receipt, and ideally, the clinical history.

User selected window and level presets for viewing different scan types should be available. Window and level should be able to be easily and readily adjustable independently of each other. The ability of the software to "remember" the last used settings is convenient if the study is later recalled and re-reviewed.

Image orientation and number of images per page need to be under radiologist control. Easy-to-use panning and zoom affecting one or all of the images is necessary.

A split screen facility enables old and new studies to be compared.

Interactive image viewing, in which what a radiologist types, draws, or points to on a workstation screen at home can be seen by a clinician on a monitor in the hospital, allows the radiologist to indicate precisely where abnormal findings are present.

Although image processing, e.g., sharpening, smoothing, etc., can help in feature extraction and identification, it introduces the possibility of removing important data as well as introducing artifacts that may subsequently be interpreted as pathology.

Cases should be capable as being marked as "reviewed" and indicated to the hospital server with the date and time stamped.

Finally, a mechanism for transmitting a report whether it be a telephone call, fax, voice dictation, or some other method must be in place. The submission of a report should be documentable.

Image Quality at Home

The on-call radiologist must receive images with sufficient detail in order to render an interpretation. If goal of home-based teleradiology is to provide a preliminary report, system requirements are less rigorous than if a final diagnosis must be made.

As already mentioned, images are compressed to speed transmission and also to save storage space. The amount of compression is inversely related to the quality of the image. That is to say, the more the image compression, the more the image degradation. The potential for removing detail by the process of compression/decompression is of concern.

Complicating the issue is that there are different compression algorithms and the same degree of compression can lead to different image quality depending on the algorithm. There are are at least two of commonly used compression methodologies: JPEG and wavelet, and for wavelet, there are at least three subtypes.

Then there is the issue of display. Plain films, or radiographs, because of their high information density require high quality monitors with 2000 - 4000 line resolution. CT and MR data, with matrices no greater than 512 x 512, have much less stringent requirements.

Many 17" monitors available today provide 1280 x 1024 pixel resolution, while 19" and 21" monitors offer up to 1600 x 1200. But there are factors other than size, such as dot pitch, screen curvature, central vs. peripheral resolution, distortion, luminance, and more, that determine monitor quality. Even an excellent monitor can produce sub-optimal results if the graphics card inside the computer cannot drive it at its maximum resolution.

Fortunately, there are alternative solutions. Flat screen technology is advancing rapidly in quality as price continues to fall. Portable (laptop) computers with 13.3" and 14.1" active matrix screens are available with large hard drives and operate at speeds that were unheard of on desktop models even a short time ago. It is arguable that for CT, MR, ultrasound and nuclear images, a modern laptop computer may very well suffice.

Specific Recommendations -- Organizational

As the technical issues of teleradiology are being sorted out, there are organizational issues to be faced as well which go well beyond radiologist staffing.

A major issue is what happens if -- or more precisely, when -- the hospital server breaks down or a radiologist's computer malfunctions. Who will fix what and when? Does the service contract include repair or replacement within 24 hours and does that apply to holidays and weekends.

Who is responsible for training of personnel and routine housekeeping chores such as backups? The responsibilities of the M.I.S. staff, hospital administration, and radiology department must be clearly understood.

So, What is Like?

So what's it like interpreting from a monitor instead of from hard copy film mounted on a viewbox? Well, for me the answer is, "Not bad at all."

First, just not having to fumble around with films, sort them, mount them on the viewbox, have them fall down the back of the viewer, only then to discover that some resident has walked off with the key images to show his Attending, is a miracle all its own. The ability to window and level CT scans just the way one likes rather than being "stuck" with what the technologists put on film is a delight.

Now, I know that I am the only one who is ever given paranasal sinus CT scans with images upside down or right to left. The ability of a computer to make everything right instantaneously is just amazing. Ditto for putting images in the center of the film and zooming it to the right size.

Teleradiology was, for me, very easy to get used to.

But not all is sweetness and light. Our home workstations are, by today's standards, slow, have small hardrives, use graphics cards that cannot take advantage of the highest resolution of our huge 21" monitors without flickering, contain internal 33.6k modems, have software that I would say, charitably, is capable of enormous improvement, and I could go on. Images are of "wet reading" or preliminary quality. The software is proprietary so that when this system goes down -- and it has -- none of my other computers can come to the rescue.

In Conclusion

Looking over my experience and that of others, I can say that teleradiology raises many issues, only some of which are technical. Technical problems are relatively easy to solve with hardware and software upgrades, especially if open standards are used.

Problems involving personnel are much more difficult to repair and are better avoided initially. Achieving consensus requires constant dialogue and open lines of communication.

Larger issues, such as the impact of teleradiology on resident training and hospital liability, also need to be examined.

Selecting and integrating new technology is a process and as such, can be standardized and critically reviewed. Many businesses are structured to follow industry-wide ISO (International Standards Organization) standards, thus assuring themselves and others that an accepted approach has been used for development and testing.

As new technologies are introduced into the hospital environment, clear procedures need to be followed in order to help avoid potential pitfalls.

Places You Might Like to Visit

The Society for Computer Applications in Radiology publish a series of informative publications including Understanding Teleradiology (1994), Understanding Compression (1997) and Understanding PACS (1992). They also maintain a (http://www.scarnet.org) and publish a journal (Journal of Digital Imaging).

An easy way to find further information is to use a search engine like Altavista and use keywords like teleradiology and DICOM. Once you find a site that seems interesting, you can follow the links that are usually contained within.

For published articles, you can do a medline search using either Internet Grateful Med or the less exhaustive PubMed, again using teleradiology and DICOM as keywords.

Newsgroup discussions can be accessed through Google Groups and use DICOM as your search word. Last time I checked there were 212 entries!

In the meantime, you can check out the following. I selected these (usually) not only for their content but their collections of links to other interesting sites.

*************

• http://www.expasy.ch/www/UIN/html1/projects/osiris/DownloadOsiris.html
• http://www.dejarnette.com/efinegan/pacspage.htm
• http://www.rasnaimaging.com/people/auro/telemed.htm
• http://web.wn.net/~usr/ricter/web/telerad.html
• http://www.rsna.org/RSNA2/practiceres/dicom.html
• http://www.erols.com/veader/
• http://www.mayo.edu/physician/mmi/teleradiology.html
• http://www.mcis.duke.edu/standards/DICOM/dicom.htm
• http://www.offis.uni-oldenburg.de/projekte/dicom/dicom_main_e.html
• http://diagrad.med.yale.edu/PACS/ARRS1/Sld032.htm
• http://wuerlim.wustl.edu/DICOM/rsna97/rsna97.html
• http://anchorage.ab.umd.edu/cgi-bin/dicomarc
• http://www.si.umich.edu/~weymouth/Search/DICOM.html
• http://www.indyrad.iupui.edu/cgi-bin/wwwdcmnew
• http://crnet4.carelian.fi/euromed/wwwdcm/wwwdcm.html

November, 1998

List of Publications -->

Return To Home -->