Computers Taking Over Rads!

[SplitBrain]

What do you think? Is this a legitimate justifiable concern in the foreseeable future (15-20 years)?

---

Members don't see this ad.

 

[medsRus]

Yes, that is precisely the future.

 

[Pilot Doc]

What do you think? Is this a legitimate justifiable concern in the foreseeable future (15-20 years)?

No.

Processing images is one of the best examples of something that people (and most vertebrates) do very well and computers do really poorly. So the likelihood of computers taking over one of the most difficult visual tasks humans do is pretty unlikely.

 

[medsRus]

No.

Processing images is one of the best examples of something that people (and most vertebrates) do very well and computers do really poorly. So the likelihood of computers taking over one of the most difficult visual tasks humans do is pretty unlikely.

Absolutely untrue. Look at automated volumetric studies in MRI.

 

[SplitBrain]

medsrus, you have no credibility whatsoever

not that I do either, because I'm in CC, and I'm an idiot, but I've come across your posts before, one second you say your about to finish med school and your POSITIVE your going into neurology then next year you say your about to go to med school. then anytime something comes up about something money you come in with "what? 1 million a year? how do i do this" so stop acting like your anything, we know your in high school

So I was asking residents, not lying high schoolers, what they thought.

 

[medsRus]

medsrus, you have no credibility whatsoever

not that I do either, because I'm in CC, and I'm an idiot, but I've come across your posts before, one second you say your about to finish med school and your POSITIVE your going into neurology then next year you say your about to go to med school. then anytime something comes up about something money you come in with "what? 1 million a year? how do i do this" so stop acting like your anything, we know your in high school

So I was asking residents, not lying high schoolers, what they thought.

Ridiculous post... I never said anything about med school. If you want a certain answer and you get the opposite from members that have been on here for years, don't bother posting your questions. I know your a pre-med and most change their minds (or don't get into med school), so addressing your concerns is not top priority. I've responses to many pertinent issues with statistics, corrected factual errors, and so on. You = zero. Bye.

 

[Taurus]

Image analysis is one of the hardest things for computers to do. A computer can crunch terabyte after terabyte of data with ease, but it can't recognize facial expressions very well at all -- something which even a 1 year old could do. Human brains are wired for image interpretation.

If computers could replace radiologists, then you will have computers replace truck drivers and airline pilots plus whatever else could be automated. Ain't going to happen in my lifetime.

 

[InternationlDoc]

By way of example newer algorithms for fully automated segmentation for HCC volume in liver are approaching 85% in sensitivity atleast in the software development in my lab. Also ICA analysis provides for one of the most robust antialiasing pure signal and thus image processing data. HOWEVER, these algorithms are in infancy. quite honestly they wouldn't replace radiologist's "clinical" knowledge, contrary, they might provide the next wave of efficiency to the field post PACS.

 

[medsRus]

Image analysis is one of the hardest things for computers to do. A computer can crunch terabyte after terabyte of data with ease, but it can't recognize facial expressions very well at all -- something which even a 1 year old could do. Human brains are wired for image interpretation.

If computers could replace radiologists, then you will have computers replace truck drivers and airline pilots plus whatever else could be automated. Ain't going to happen in my lifetime.

Most commercial airplanes have automated flight. Also, there are systems in place for automated driving (and betas in several cities).

 

[bulletproof]

Most commercial airplanes have automated flight. Also, there are systems in place for automated driving (and betas in several cities).

Yes, and yet every commercial airplane still has at the very least one pilot in the cockpit. Not to mention the fact that automated systems are disengaged for both take off and landing, you know, the most dangerous portions of flight. They are routinely disengaged in inclement weather etc. I don't foresee any plans to get rid of pilots in commercial airplanes anytime soon. I predict that when that day of reckoning arrives, Radiologists will be at least 3% closer to being replaced by computers.

 

[Faebinder]

You all act like radiology is a cookie cut science. I can't count how many times the radiologists were unsure if the pons/medulla lesion was called an artifact instead of an infarct. At a certain point someone with a license will need to take the risk and make the call whether it's a donkey or a horse.

In addition, radiology is advancing faster than computers can catch up. We dont even have a program to read the simple chest xrays and yet there are fMRIs, SPECT scans, PET scan and all sorta new imaging modalities that we barely scratched their surface of use.

This thread needs a due to sheer lameness.

 

[Taurus]

Most commercial airplanes have automated flight. Also, there are systems in place for automated driving (and betas in several cities).

But how is the data being fed into autopilots? It's GPS. The autopilots are not interpreting the visual cues in real-time to fly or drive.

CMU has the most advanced auto-driving system around, but even they say it is far from being read commercially ready. If you ask any computer science professor, image interpretation like artificial intelligence is one of the hardest things to accomplish in their field.

 

[Neuronix]

This thread needs a due to sheer lameness.

Lameness and the fact that it's a duplicate thread.

http://forums.studentdoctor.net/showthread.php?t=652545