The field of Radiology will be replaced by advanced computers

[DrGregoryHouse]

My opinion: AI will not take over Radiology from one to the next, but, rather, computers will at first work with radiologists and then slowly take over more tasks which they are able todo more efficiently. At first AI programs will make their impact in small domains (such as now in mamorgaphy and chest CT scans) where the images have rather constrained statistics. Right now companies like R2 and such suggest the computer will act as a second reader, working in conjuction with radiologists. But as these programs become more efficient, they will become more reliable and ultimately supplant radiologists (again only in certain domains). A general-purpose AI reader is far off in the future, maybe 30-40 years.

my 2 cents ...

I agree.

---

Members don't see this ad.

 

[durfen]

One way to escape being replaced is to do the replacing. My work during undergrad and med school pretty much revolves around computer aided interpretation and diagnosis.
However, as definite as the first sentence sounds, I am confident that the importance of imaging will exponentially increase over the next few decades and we will NEED computers to do some of the work for us in order to keep up with the demand.

 

[Arsenic]

IMHO if there's a field that can be replaced by computers it's IM... let me think back to my high school computer math days...

10 enter patient
20 symptoms = PATTERN RECOGNITION
30 treatment = ALGORITHM MEMORIZATION
40 discharge
50 if patient deteriorates and returns with same symptoms, GOTO 10
60 if patient dead, END

My 5-Minute Clinical Consult is a fantastic Internist... much better than I am anyway...

damn, I'm screwed!

 

[bigfrank]

Good thread!

I'm gonna go put my head in the oven now.

 

[Strength&Speed]

Im not a radiologist...but the point about mammography is a good one. A computer can't compare prior studies, see what's benign from not benign...there are just too many variables. Most things in medicine are contextual. I try to make this point with my colleagues. A tsh of 7.35 and a free t4 of .67 is probably subclinical hypothyroidism....not myxedema coma even if the clinical presentation might appear as such. Hard numbers are interpreted differently depending on the clinical presentation...which can vary in literally innumerable ways. For instance...take a BNP of 600....does this patient have CHF? well...maybe, have they had recent heart strain? heart attack? tachycardia? decreased renal output? pneumonia? pe? obstructive sleep apnea? depending on the severity of any of the previous, they may or may not have chf according to the BNP. And you have to grade and judge the severity. Its not just looking at a lab or an image, and saying..."oh, it's this" Some of the time...probably most of the time..."it depends".

As another example...I had a case of a woman who had a 3x3 cm "mass" in her pelvis who was treated with antibiotics for 10 days. Turns out after reviewing the film it was her left ovary. Only after comparing the film with a previous one and seeing the slightly posteriorly placed left ovary could they make that call. I doubt you'd design a computer program to figure that out.