I think VR is a useful technology. Ability to place phrases into reports that you always say shortens time dictating (use of macros). Use of speech shortcuts (i.e. train vocabulary lexicon to say what you want by saying acronyms) helps a lot.

Dragon out of the box allows a lot more flexibility in placement of cursor, moving around the page, selecting words, formatting, etc, and even creating custom commands (like tab forward three times, or go to line 25, etc). The radiology versions of software are extremely top heavy and limiting and every time I have pointed out either bugs or design flaws, I get told that the program is supposed to work that way.

It is frustrating especially when I know that there is better stuff out there.

And some of the companies will automatically take out all "and"and "the" as a rule of thumb, as if the programmers decided those are nonsense words (like uh or er). It changes the meaning of sentences and the companies won't do a whole lot about it.

One funny thing here--there is a typo above. Should say "voice recognition"but it says "voice regnition". I wonder if that was typed or voice transcribed?? LOL

VR is potentially an extremely useful tool to increase practice efficiency and report turnaround. It is however unfortunate that the technology is almost completely monopolized by a company that is one of the worst for customer service I have ever dealt with. All our practices are different. The only answer you are likely to hear from them is "if you are unhappy with the features you will have to purchase a newer version". There is no flexibility and therefore your workflow is 'dictated' by the limitations of their software. I wish other major vendors would come in and put some pressure on them to proved better service.

Voice recognition, unfortunately, doesn't take "instructional"dictation, e.g., "Start with a standard normal, go to the part that says abdomen and pelvis, change that to chest abdomen and pelvis. Also, be sure to send a copy to Dr. Smith. Go back up to where I said T12; change that to read T11. Please correct the comparison date to 9/11/2011, not 12/7/2011. Centerlake is spelled CentreLake. Add this reference at the bottom, below my signature: Radiology, Volume 334, 2011. change the contrast material dose to read 60 mL instead of 100mL."

Voice recognition requires shifting gaze and visual attention from image to text to image to keyboard to text to keyboard(multiple nonsimultatneous visual tasks), rather than allowing simultaneous speech and vision, neither interruptinig nor distracting nor degrading the other

Voice recognition is fine for short, standard, near normal reports. Complex multicomparison multimodality examinations, such as a series of PET-CTs and MRs compared over several years, require an essay of a full page or more: voice-typing, keyboarding, and proofreading are not efficient uses of a radiologist' visual workload or attention.

Voice recognition is poor at creating attractive and clarifying formatting, such as indents, tables, lists, bold, italics, footnotes, and numbering.

I've found that the best of all worlds is a transcriber who likes what she does, knows her radiologists' work and word style, and stays ten minutes behind his microphone. I've been fortunate to have those for decades, and/or to train them to be such. The labor cost of this lady, of course, is not insubstantial, but the quality and physician efficiency is "worth it," depending upon what employer is able to keep her gainfully employed in this jobless recovery.

While some of my best friends are machines, there are many things humans would not have computers do.

I absolutely concur with Johnson Lightfoote. I have used VR on and off at various times over the past 12 years. I still find it error prone, inflexible and cumbersome. No software can take the place of a really good transcriptionist. Editing text is an extremely poor and inefficient use of a radiologist's time. I hate VR and I don't use it if I can help it. Having said that, there is a caveat: if you're working in an outpatient setting, it's easier to use transcriptionists than if you're working in a hospital setting where there are never ending streams of stats. VR has the advantage there due to its inherent speed edge, assuming that the report isn't too long and doesn't take too long to edit if it needs editing. You'd have to have a pretty darn crack transcription pool to keep up with the torrent of stats from a busy ER.

I would prefer to not use voice recognition. I use it only because the administration insistsand no longer has transcribers and refuses to hire proof readers. I find it error prone and very difficult.

I have used voice recognition software for years.It remains an excellent Beta version and even a pretty good version 1.0 or 2.0. But after more than 10 years it has not significantly improved. It is stuck in mediocrity.

Changes that would vastly improve accuracy and reduce stupid errors are easy to make, yet the hospitals and the software companies seem disinterested in suggestions and the software remains mired in, "well it does get a report out faster."

Improvements are easy. But only if the companies listen and weigh the value of suggestions. This has not happened.

Note: this blog was typed by hand.

Alan, I wish I could blame VR software for the typo. That's human error! The mistake has been fixed.