CAD's chances may be remote

February 25, 2004

By Greg Freiherr, Editor, gfreiherr@cmp.comPeople love machines, especially ones that make their lives easier. I, for one, love remote control. I use the devices for every

By Greg Freiherr, Editor, gfreiherr@cmp.com

People love machines, especially ones that make their lives easier. I, for one, love remote control. I use the devices for every form of electronic entertainment in my house-television, DVD, VCR, stereo. But I only want them to do so much. I would draw the line at allowing them to replay scenes that that I might have missed or that might benefit by a different angle.

Oddly, I don't mind people doing this. For example, I'm a big fan of John Madden scrawling on football instant replays on my television screen. I first discovered this preference for flesh and blood analysts, however, not in my living room, but at the University of Pittsburgh about 25 years ago. At the time, computer jockeys were trying to build what they euphemistically called "Internist." But M.D.s weren't so crazy about a silicon-based wannabe expert of internal medicine. Diagnosis, they told me, should be left to real doctors.

In the end, Internist didn't get very far. One reason was the impossibly complex task of programming all the knowledge of internal medicine. Since then, however, more manageable tasks have been chosen, data processing has gotten faster, and programmers have gotten better. One result-computer-aided detection-seems to have captured the fancy of many in the radiology community. Having established a foothold in mammography, it may branch out into other modalities, notably chest CT.

The reason seems obvious enough. Radiologists are having a hard time keeping up with the volume of data being generated by CT scanners. The problem is only going to get worse, as more powerful CTs generate ever more data. But I think it is a mistake to believe that CAD is the only answer-or even the preferred answer. It's possible that the popularity of CAD has grown at least partly because of the current shortage of radiologists. And that might be coming to an end.

Research conducted by the American College of Radiology, published in this month's American Journal of Roentgenology, suggests a dramatic decrease in the radiologist shortage in the U.S. Most intriguing was a shift in attitudes. In 2000, 51% of those surveyed reported too much work with 5% reporting too little. Last year, only 17% said they were overworked and-here's the zinger-16% indicated a desire for more work.

In the future, if this trend continues, CAD might be seen as taking work away from radiologists. If that happens, CAD's prospects may be no brighter in radiology than on "Monday Night Football."