how long before computers can interpret CT/Xray/MRI?

[MacGyver]

i know that we dont yet have this capability, but i have to think that its only a matter of time before computers will become sophisticated enough to do all the interpretive work.

There are good fuzzy logic software cores beign applied to other problems that dont have mathematical consistency, this seems to be an ideal way to approach radiograph interpretation.

predictions or thoughts?

---

Members don't see this ad.

 

[MustafaMond]

Your are ignorant of current trends in radiology.
See this thread
http://www.studentdoctor.net/forums/showthread.php?s=&threadid=42153&pagenumber=1

Sorry to be mean, but I never liked your posts from before, Mac. (Everyone Forum etc.)

The word "know" is used waaaaay too often by you and, as usual, it exposes your ignorance.

 

[xraydoc]

The closest thing to this idea is the R2 CAD in mammography. I use R2 at least once a week and it pretty much sucks. If for example you were to just let R2 CAD "intrepret" the mammos you would biopsy about 75-90% of all women and nearly all of the biopsies would be negative.

Like many fields of advanced study, Radiology IMO is an inexact ART not a science in the strict sense of the word. Computers would have a very difficult time simulating subjective experiential things such as MRI or CT. This will not happen in my practice life time (25 years) other than perhaps a type of CAD to look for lung nodules, lymph nodes etc, but the radiologist will provide the bulk of the brain power.

The only people who I hear talking about computers replacing radiologists are computer people who know near nothing about what it takes to look at a complex study such as a MRI or CT of the body.

 

[droliver]

I don't know about predictions with this. The computer tech improves so very rapidly I would be surprised if it was really 25 years before we had some working software for this used regularly. God knows how it would be integrated into practice - a screening tool perhaps with human oversight?

 

[javanomad]

Image Recognition is just not comming on as fast as some other computer technologies. Actually IR is a very tough problem and only Neural Networks have come close in showing a solution.

IR in it's current state is really a software coding problem more than in speed or hardware. Today I would say we have the computing power to do medical grade IR but the software or "rules" are sorely lacking.

That being said I think in the next 5-10 years we WILL see more IR pre-processing of films. The current workload of Rads makes this inevitable. After all if the Rad can run through 2x films a day using some techno widget some med admin wonk will buy the techno widget.

Of course since people can't sue the computer...

Now if we could just get some technology to ward off the lawyers... then we would have something!

 

[Al Pacino]

I think a even bigger concern is the possibiliy that radiology reads will be outsourced to foreign radiologists who will do the reads at a cheaper price. In fact, MGH has a link on their website about its possible collaboration with Indian radiologists who will read the films for cheap. Mass Gen's Imaging Homepage


So the demise for US radiology may not come from computers who can read film-- it'll probably come from those who are willing to do the reads for a very low price.

 

[southerndoc]

I hear Sudan is running a blue light special on rad reads this month.

 

[Ctrhu]

MGH's plan doesn't make sense. There is not enough residency positions for all the US applicants. And they want to bring Indian physicians here to train and then send them back? To practice radiology in the states, a physician must have all the credentials. I assume that these Indian docs must also. Why would such doctors go back to Inida to make a fifth of what they can earn over here? Foreign MDs have been coming to the US for training for years, but MOST elect to stay.
There are groups in Australia that are recruiting US rads to move there and provide after hours coverage for US facilities taking advantage of the reverse day/night.
I believe telerad will never take the place of a local rad because of the interactions between the radiologist and the referring physicians and with the patient. I cover four facilities, two of which are by telerad and this arrangement has been unsatisfactory as I tailor the protocols for each patient and spend a lot of time answering questions from referring physicians. This exchange is vital. Referring docs can also be very demanding - they want reports ASAP and would have trouble dealing with physicians half a world away with difficult to understand English.
I would not be worried if I was considering to specialize in Radiology.