The way radiology reports are generated has certainly changed since I began my radiology residency in the Navy in the late ‘70s. We dictated on strange little belt dictation machines that resembled something Thomas Edison might have used. These belts disappeared into the bowels of the hospital and 10 weeks later (yes, I said 10 weeks) the reports reappeared for our review and signature.
At one point we were ordered to stop dictating and handwrite all reports because of the delays. Finally, someone had the brilliant idea of shipping our little belts to the other side of the country and the turnaround time was reduced to two weeks.
Eventually, we got cassette tapes and the transcription department of the hospital started to improve. The hospital had a central transcription pool. Apparently, transcribing radiology reports wasn’t popular in the pool, possibly because of the large volumes of work and words. The task of transcribing them fell to a few unlucky souls, especially poor Gertrude. One of her most infamous works was a lengthy arteriogram report. It appeared that all of her fingers were positioned one key to the right on the keyboard and the entire report was gibberish. The sad part was no one complained.
Later in my naval career, a radiology information system (RIS) was installed that included a dictation module. Since this still relied on human transcriptionists the improvement wasn’t all that great. The system did have a feature that allowed for barcoding normal reports, however. The system was designed for an entire report to be generated by swiping a barcode, but a buddy of mine and I discovered that the barcode reports could be “stacked” and we created binders of barcodes that allowed us to create a variety of normal and routinely abnormally reports by swiping several barcodes one after the other. This bypassed the transcriptionist and greatly improved our report turnaround time.
After the Navy in various civilian institutions there followed a variety of dial-in and digital transcription services, and eventually we were gifted or punished with voice recognition (VR), depending on your point of view. This amazing technology transforms the spoken word into the written word before your very eyes. Never mind that the work of the transcriptionist is largely shifted to the physician.
The ability to dictate final reports on call obviating the need for scribbled handwritten reports and reading the same study twice (once on call and once the next day) was definitely a step in the right direction. Provided, of course, you could get the software to understand you. “Probably” comes out “bubbly bleed,” “extremity” comes out “Germany,” “paresis” comes out “power assist,” “portable” comes out “herbal,” and the ultimate challenge — the number 2 can come out as “2,” “two,” “to,” or “too” depending on the machine’s mood. I get a kick out of the spellchecker on VR. Why it produces a word then questions the spelling of the word it just produced is beyond my comprehension.
The words it never seems to get wrong are the expletives that are frequently yelled into the microphone by my frustrated fellow rads trying to be understood. I guess the programmers were a bawdy group after all, possibly old sailors.
