Voice recognition software in radiology?

[GH253]

Would it be possible to utilize voice recognition (i.e. speech to text) software in radiology when writing reports? Or would this be too impractical?

---

Members don't see this ad.

 

[MossPoh]

Many places already do use it...

 

[colbgw02]

I believe the first department to install voice recognition software was back in the 1996-97 time period. It was a government facility that was used as the manufacturer's beta test site, so they got the software for free. As already mentioned, it is quickly becoming commonplace. I know that I would not even consider a job that did not include dictation software and a PACS, and I imagine that the vast majority of recent and current residents feel the same.

 

[Winged Scapula]

Many of the reports I receive have some little note that they were dictated using such software.

I can tell which radiologists and pathologists actually proofread their reports, because these systems are not foolproof, even if you train them.

I use Dragon Speak in my office and some of the things its types are fairly funny or nonsensical. Fortunately, its not common so its much more efficient and cheaper for me to do it myself than use a dictation/transcription service.

 

[Twitch]

It'd seem that most people these days can type faster. Despite advances in voice recognition you'll still see folks go in and correct a word or two it after a certain number of lines. Just seems like a lot of hassle.

 

[ioncell]

Many of the reports I receive have some little note that they were dictated using such software.

I can tell which radiologists and pathologists actually proofread their reports, because these systems are not foolproof, even if you train them.

I use Dragon Speak in my office and some of the things its types are fairly funny or nonsensical. Fortunately, its not common so its much more efficient and cheaper for me to do it myself than use a dictation/transcription service.

It's nice to have Dr.Cox here


These voice softwares are crap until this moment .

my 2 cents.

 

[colbgw02]

It'd seem that most people these days can type faster. Despite advances in voice recognition you'll still see folks go in and correct a word or two it after a certain number of lines. Just seems like a lot of hassle.

I'll admit that I will routinely use the keyboard to correct minor errors in my dictations, especially when I know that it involves a word or phrase that the software slips up on. However, I would never consider just flat out typing a report.

First of all, many studies are normal or have only slight abnormalities, which allows me to use a canned text with little or no modification. I can bring up paragraphs of pre-determined dictations with only a few words. This feature was obviously available before software, but now they can be personalized and constantly updated to a greater degree.

Secondly, for a complex study, I think most people prefer to dictate as they go, and the software allows for betting editing and arrangement of the information than traditional dictation.

Lastly, it essentially eliminates the cost and hassle of transcriptionists. I personally find it annoying to have to proofread a report hours or days after dictation was actually done. There's almost always too much volume to remember a specific case, so when the transcriptionist types something nonsensical it requires a lot of extra time to revisit the images to determine the correct phrase.

 

[MossPoh]

It'd seem that most people these days can type faster. Despite advances in voice recognition you'll still see folks go in and correct a word or two it after a certain number of lines. Just seems like a lot of hassle.

As was noted above, the software allows a lot of things. It is by no means perfect but I've seen the early versions relative to the new ones and I can say that every year it gets better. Many people CAN'T type faster than they can talk. I can, but my dad is a hunt and peck type guy. It makes editing pretty slow with him. Also, spelling is a pain for quite a large number of people. Hell, even with basic words I catch myself sounding it out in my head once in a while. It would be insanity for some places to type reports. The "average" day at the hospital I did my analysis was still 110 cases per radiologist. You're just asking for all the issues from crap ergonomics to type even a fraction of that.

I actually did a usability analysis of this as part of a thesis and it does turn out to be more efficient to use the voice recognition software IF the person (and software) is properly trained. Mumblers have a tough time and have to alter their speech patterns.

Just the ability to have the endless amount of macros set up is great and can be a real time saver.

 

[kingalls]

Amen - our facility does quite a bit of telerad, often using different dictation systems depending on the facility we're reading for. We have transcription, powerscribe, dragon, and good old fashioned typing. Normal or not, you're begging for carpal tunnel if you're typing 180 reports/shift. Macros and Templates are godsends.

 

[Winged Scapula]

Between templates and Dragon Speak I have cut down my chart time by 80%.

I can type fast but the above make me much more efficient, even if I do have to go back and make a few corrections.

 

[Lacuna]

Speech recognition is more efficient but prone to errors. I pulled these articles from AM to show that:

Report turnaround time in one institution went from 28 hours to 12.7 hours following implementation of speech recognition.

http://www.auntminnie.com/index.asp?Sec=sup&Sub=ris&Pag=dis&ItemId=86233

However, in a study of 256 CT reports, 22% had dictating errors that were classified as significant or confusing.

http://www.auntminnie.com/index.asp?Sec=sup&Sub=ris&Pag=dis&ItemId=84270

How many of you review the report before submitting it? How long does it take to review?

 

[kingalls]

Depends -when using something like Powerscribe or Dragon and the words are right there in front of you, reviewing it isn't much trouble. The problem comes when you're using transcription, and sometimes several hours elapse between dictation and review.

 

[cleothecat]

Quick turnaround time

Spend more time finding "errors" in my report, rather than pathology on Images.

 

[Angio]

I am not a Rad, as I am pre-med, however I do use Dragon Naturally Speaking to write all of my essays and reports for all of my classes. I certainly cannot type at 100+WPM... besides, the accuracy for me so far has been in the 98% range.

At this rate... It seems as though transcriptionists are on their way out.