The end of Radiology?

[DrJD]

http://www.npr.org/2011/02/20/133916058/the-dark-side-of-watson?ft=1&f=1001

Happened across this article, and was surprised to see that a "reputable" news source was reporting on it.

Please know I am not asserting what the author is, but I thought it would be good for everyone here to see what information is going out to the general public. What do you guys think?

---

Members don't see this ad.

 

[quietude]

http://www.npr.org/2011/02/20/133916058/the-dark-side-of-watson?ft=1&f=1001

Happened across this article, and was surprised to see that a "reputable" news source was reporting on it.

Please know I am not asserting what the author is, but I thought it would be good for everyone here to see what information is going out to the general public. What do you guys think?

Sounds like they picked someone who hasn't personally worked on CAD image processing. I am sure that somewhere down the road our jobs might be in jeopardy, but this is somewhat like fretting that one day we won't drive our own cars or I, Robot will stop being a futuristic story.

Based on the journal articles I've seen about image processing software, I am not worried. You might as well argue that doctors will be replaced by algorithmic decision-making machines guided by a PA-type person who does the physical exam and the labwork.

The human factor will win out in our lifetimes. If you are really that concerned, pick mammo, IR or MSK and make a living out of procedural-based radiology.

 

[badasshairday]

IBM is the real life version of "Skynet" from the Terminator series. The machines are taking over!

 

[DrVanNostran]

IBM is the real life version of "Skynet" from the Terminator series. The machines are taking over!

Picking Radiology as a career is scaring the crap out of me. I think I need to re-organize my 4th year schedule

 

[badasshairday]

Picking Radiology as a career is scaring the crap out of me. I think I need to re-organize my 4th year schedule

In the short term, the rise of the machines will be great for Radiology. Think about it, with the aid of a prelim reading by the machine, Radiologists will be able to read studies even faster. They will have to be their to make sure the machine did not screw up. More studies read equals higher income. Although the higher volume will be undercut by cuts to imaging reimbursements. But Radiology will continue to one of the best paid fields with the advancement of technology.... until the day the MACHINES TAKE OVER!

I'd say you have a good 20-30 year career before that. After that, we are all ****ed. Surgery will be done by robots, Internal medicine by machines with input by a technician. But whatever, by then the world will be like the movie Idiocracy. LOL, just a light hearted post.

 

[C5toC9]

I think it's similar to EKGs given prelim readings by the EKG machines. Cardiologists always look over their own EKGs because the machines can mess up.

 

[colbgw02]

If the current CAD used in mammography is any indication of how well computers do radiology, then I think our jobs are pretty safe for ... Oh...the next 50 to 100 years,

 

[DrVanNostran]

In the short term, the rise of the machines will be great for Radiology. Think about it, with the aid of a prelim reading by the machine, Radiologists will be able to read studies even faster. They will have to be their to make sure the machine did not screw up. More studies read equals higher income. Although the higher volume will be undercut by cuts to imaging reimbursements. But Radiology will continue to one of the best paid fields with the advancement of technology.... until the day the MACHINES TAKE OVER!

I'd say you have a good 20-30 year career before that. After that, we are all ****ed. Surgery will be done by robots, Internal medicine by machines with input by a technician. But whatever, by then the world will be like the movie Idiocracy. LOL, just a light hearted post.

If the current CAD used in mammography is any indication of how well computers do radiology, then I think our jobs are pretty safe for ... Oh...the next 50 to 100 years,

Great points by both of you.

I was just going to mention CAD. I have been working at a Women's imaging station during my rotations, and 90% of time it just picks up stroma. While this gives me confidence in our field, it makes me worry about when we do start relying on machines lol.

 

[Taurus]

Having degrees in both computer science and computer engineering plus years of experience in large-scale software development and now being in radiology, I think that I am pretty qualified to give a response. The rest of medicine will be replaced by Watson-type programs way before radiology ever will. Image interpretation is extremely difficult for computers, probably the most difficult problem for it to solve. I feel confident that computers won't replace me in my working lifetime or even in a 100 years. If the day comes when computers can replace radiologists, I think mankind has needs to worry about its own existence. A sentient computer wouldn't be far off. The guy who NPR quoted as saying that radiology is in danger is "software programmer and a writer". He's not a physician and he's certainly not a radiologist. He's clueless.

 

[PokerDoc]

With all due respect to Mr. Ford, he comes across as incredibly arrogant to me in that statement. I don't think he has any clue as to the complexity of diagnostic imaging in medicine to be so bold and assert that machines are anywhere even close to achieving that capability.

 

[DrRads101]

Lol. Sorry but if computers can't even get ekg's right most of the time, from personal experience, how the heck are they going to tackle Rads. It just strikes me as someone who has not actually sat with a Radiologist before and witnessed the plethora of studies and findings that come up.

 

[cowme]

http://www.npr.org/2011/02/20/133916058/the-dark-side-of-watson?ft=1&f=1001

Happened across this article, and was surprised to see that a "reputable" news source was reporting on it.

Please know I am not asserting what the author is, but I thought it would be good for everyone here to see what information is going out to the general public. What do you guys think?

 

[DrJD]

Since Watson just made the Jeopardy debut recently I doubt this thread has been posted before ;-)

 

[DrJD]

Having degrees in both computer science and computer engineering plus years of experience in large-scale software development and now being in radiology, I think that I am pretty qualified to give a response. The rest of medicine will be replaced by Watson-type programs way before radiology ever will. Image interpretation is extremely difficult for computers, probably the most difficult problem for it to solve. I feel confident that computers won't replace me in my working lifetime or even in a 100 years. If the day comes when computers can replace radiologists, I think mankind has needs to worry about its own existence. A sentient computer wouldn't be far off. The guy who NPR quoted as saying that radiology is in danger is "software programmer and a writer". He's not a physician and he's certainly not a radiologist. He's clueless.

Thank you for sharing your experience Taurus. Relatively few people have the relevant education and experience in both computer science and radiology to weigh in effectively.

Personally the article seemed at the very least far fetched. From my unexperienced perspective it seems like a Watson like computer would much sooner be able to perform a simple DDx in the realm of Family medicine, etc than read studies performed in radiology.

 

[Taurus]

Waton-type computer won't replace radiologists but I predict that it will be a huge benefit to radiology and most of medicine.

I believe that there is a huge opportunity to create the equivalent of "google" for medicine. I want a technology that reviews all prior notes, labs, radiology, etc and spits out an executive summary (accurate and timely medical and surgical history, relevant clinical info, etc). Because radiologists don't see patients, we have no clue about the medical history. For example, we may see "pain" as the indicator for an x-ray. We have no idea that this person has positive lab results for rheumatoid, gout, etc. Because of the high volumes these days, we don't have much time to look through pages and pages of notes about each patient to find that moment of "Aha! He had a right hemicolectomy in 1965 and that's why his colon is so funky looking!". Radiologists have to pick and choose when to spend the extra time and effort to look up information. What I want is to get a detailed but pertinent summary of the current medical condition of a patient so that I will know nearly as much about a patient as his PCP in 30 seconds. If Watson-type program or even google can make this happen, this will tremendously help radiology. Right now, the biggest weakness of radiology -- and one way other non-radiology fields are taking advantage of radiology -- is that we don't see the patient and therefore we don't know the clinical information (unless we spend the extra time and effort to do it).

 

[PistolPete]

I want a technology that reviews all prior notes, labs, radiology, etc and spits out an executive summary (accurate and timely medical and surgical history, relevant clinical info, etc).

Great idea. Next person to do this would make some serious $$$.

 

[fun8stuff]

http://www.npr.org/2011/02/20/133916058/the-dark-side-of-watson?ft=1&f=1001

Happened across this article, and was surprised to see that a "reputable" news source was reporting on it.

Please know I am not asserting what the author is, but I thought it would be good for everyone here to see what information is going out to the general public. What do you guys think?

I would be more concerned about the photo recognition technology that is similar to what is used in the Amazon iphone app that allows the user to take a picture of something and then it searches amazon for something similar. From my experience, this works pretty well for common items. I can imagine someone creating a huge data base of normals for say every slice of a body CT (or organ) or head CT/MRI and then using the computer to compare each of the patient's slices to the normal database. The computer would then mark areas of abnormality and draw the radiologist's attention to these areas.

Eventually, you could have a database of abnormals specific to the area of where it finds the abnormality. For example- abnormality in the RLQ in area of appendix compares the patient's images to 1000 varieties/complications of appendicitis, cancer, backwash ileitis, etc... Abnormality in the pancreas- compares with database on pancreatitis, cancers, etc. Clinician suspects stone? The new renal protocol searches the area of the kidneys and ureters for a high density stone. Eventually, the computer could use key patient information (symptoms, age, vitals, labs) to narrow the differential and help focus its search. Of course, I think it would have a lot of problems for someone who has really messed up anatomy...i.e. whipple procedure or someone with mass trauma, but I think specific databases could be established for the different scenarios.

Even with something similar to this, I think there will be an intermediate phase where a radiologist is still absolutely essential. It is really hard to imagine a computer replacing us in our life time.

 

[p53]

I would be more concerned about the photo recognition technology that is similar to what is used in the Amazon iphone app that allows the user to take a picture of something and then it searches amazon for something similar. From my experience, this works pretty well for common items. I can imagine someone creating a huge data base of normals for say every slice of a body CT (or organ) or head CT/MRI and then using the computer to compare each of the patient's slices to the normal database. The computer would then mark areas of abnormality and draw the radiologist's attention to these areas.

Eventually, you could have a database of abnormals specific to the area of where it finds the abnormality. For example- abnormality in the RLQ in area of appendix compares the patient's images to 1000 varieties/complications of appendicitis, cancer, backwash ileitis, etc... Abnormality in the pancreas- compares with database on pancreatitis, cancers, etc. Clinician suspects stone? The new renal protocol searches the area of the kidneys and ureters for a high density stone. Eventually, the computer could use key patient information (symptoms, age, vitals, labs) to narrow the differential and help focus its search. Of course, I think it would have a lot of problems for someone who has really messed up anatomy...i.e. whipple procedure or someone with mass trauma, but I think specific databases could be established for the different scenarios.

Even with something similar to this, I think there will be an intermediate phase where a radiologist is still absolutely essential. It is really hard to imagine a computer replacing us in our life time.

Not only is this possible..it is probable. Why? Money. The group of radiologists that develop this will all become billionaires. Mark my words. Bright minds have been going into rads the last ten years. They can read the writing on the wall for the future of radiology. The next Mark Zuckerberg/Bill Gates billionairr will be a radiologist that sells himself out to automate all CTs.

 

[badasshairday]

p53, why do you hate radiology so much now? I was looking through some of your old posts, and you loved it before. Not so much anymore.

 

[The Dark Knight]

In the short term, the rise of the machines will be great for Radiology. Think about it, with the aid of a prelim reading by the machine, Radiologists will be able to read studies even faster. They will have to be their to make sure the machine did not screw up. More studies read equals higher income. Although the higher volume will be undercut by cuts to imaging reimbursements. But Radiology will continue to one of the best paid fields with the advancement of technology.... until the day the MACHINES TAKE OVER!

I'd say you have a good 20-30 year career before that. After that, we are all ****ed. Surgery will be done by robots, Internal medicine by machines with input by a technician. But whatever, by then the world will be like the movie Idiocracy. LOL, just a light hearted post.

That won't change the number of people that will need radiographic imaging done. If a radiologist can read more images, we probably will end up needing far less radiologist. The economics will work itself out where those radiologist that do still hold on to a job may get a bump up in their salary but there will be alot of places down-sizing their workforce if that occurs. I know this isn't as rosy as a picture as you would like but the fact is im agreeing with you just showing you the not so pleasant side of your prediction. Doctors and patients are both getting wiser about unneccesary imaging and the exposure associated with it so don't think just b/c you can read faster we'll be doing more imaging. Im no radiologist but I don't think there's any issue currently with being too few radiologist to read enough images. Net effect seems like there would be a good number of layoffs and the guys left with a job would make more.

 

[copacetic]

having worked with engineers desinging image processing and interpretation software, there is no chance computers will be able to do what radiologists do anytime soon. The challenges are ridiculous. Not for another 60-100 years really will they be able to do anything even half ass decent. Not saying it wont happen, but trust me, not anytime soon.

 

[theRainmaker78]

I think it's similar to EKGs given prelim readings by the EKG machines. Cardiologists always look over their own EKGs because the machines can mess up.

to quote an attending, the worst medical student can read EKG's better than the computer. Its rather impossible to write a code so complex that it could come even close to matching the processing and calculations that the human mind does when it reads EKG's or for that matter, interprets Xray's. Technology will only make the lives of radiologists easier.

So calm down. The only thing you have to fear is outsourcing

 

[copacetic]

to quote an attending, the worst medical student can read EKG's better than the computer. Its rather impossible to write a code so complex that it could come even close to matching the processing and calculations that the human mind does when it reads EKG's or for that matter, interprets Xray's. Technology will only make the lives of radiologists easier.

So calm down. The only thing you have to fear is outsourcing

and outsourcing is unlikely to happen as well due to the legal ramifications.

 

[hartbot]

So is the future bright and sunny?

 

[theRainmaker78]

and outsourcing is unlikely to happen as well due to the legal ramifications.

sadly, I have heard of it happening in the US. All you need is one radiologist in the US to sign off on the report written by whoever (from India).
I also heard there were tons of errors made in these reports but hospital saves money so...

 

[copacetic]

sadly, I have heard of it happening in the US. All you need is one radiologist in the US to sign off on the report written by whoever (from India).
I also heard there were tons of errors made in these reports but hospital saves money so...

wait till somebody gets sued big time

 

[rocksalt]

Well! You'll be happy to know that Amazon's Image Recognition System AKA Mechanical Turk is actually large scale crowd-sourcing.

I would be more concerned about the photo recognition technology that is similar to what is used in the Amazon iphone app that allows the user to take a picture of something and then it searches amazon for something similar. From my experience, this works pretty well for common items. I can imagine someone creating a huge data base of normals for say every slice of a body CT (or organ) or head CT/MRI and then using the computer to compare each of the patient's slices to the normal database. The computer would then mark areas of abnormality and draw the radiologist's attention to these areas.

Eventually, you could have a database of abnormals specific to the area of where it finds the abnormality. For example- abnormality in the RLQ in area of appendix compares the patient's images to 1000 varieties/complications of appendicitis, cancer, backwash ileitis, etc... Abnormality in the pancreas- compares with database on pancreatitis, cancers, etc. Clinician suspects stone? The new renal protocol searches the area of the kidneys and ureters for a high density stone. Eventually, the computer could use key patient information (symptoms, age, vitals, labs) to narrow the differential and help focus its search. Of course, I think it would have a lot of problems for someone who has really messed up anatomy...i.e. whipple procedure or someone with mass trauma, but I think specific databases could be established for the different scenarios.

Even with something similar to this, I think there will be an intermediate phase where a radiologist is still absolutely essential. It is really hard to imagine a computer replacing us in our life time.

 

[fun8stuff]

Well! You'll be happy to know that Amazon's Image Recognition System AKA Mechanical Turk is actually large scale crowd-sourcing.

Interesting.

 

[ashkenazi1234]

why so many unfilled radiology spots (41) this year? end of radiology?

 

[The Dark Knight]

having worked with engineers desinging image processing and interpretation software, there is no chance computers will be able to do what radiologists do anytime soon. The challenges are ridiculous. Not for another 60-100 years really will they be able to do anything even half ass decent. Not saying it wont happen, but trust me, not anytime soon.

Lol. 15 years ago a few people would carry pagers to get in touch with someone not near a phone. Today a device that can fit in your pocket lets you make phone calls, browse the web, check email and play video games with graphics equivalent to a super nintendo..

but you think it will be at least 2071 before a computer will be able to pattern match images. im not going to argue with that. if you're that confident in your job security, i don't have any evidence to disagree with you

 

[The Dark Knight]

wait till somebody gets sued big time

Exactly. I still agree you will need HUMAN radiologist. Just not as many. That was the point I was making with the previous poster. He was saying that with these computers making an "initial" interp it will let the radiologist focus his eyes at the problem being mentioned and evaluate it faster. Although that shouldn't be the case b/c a human radiologist should look at every dam thing on that image even if its not mentioned by the computer, i figured id give this guy the benefit of the doubt. At the end of the day if a radiologist can get through more images faster, than there will be less radiologist because there sure as hell won't be more IM or surgery doctors order CTs just because the scanner is open right now

 

[DrVanNostran]

why so many unfilled radiology spots (41) this year? end of radiology?

Bump!

 

[fun8stuff]

Exactly. I still agree you will need HUMAN radiologist. Just not as many. That was the point I was making with the previous poster. He was saying that with these computers making an "initial" interp it will let the radiologist focus his eyes at the problem being mentioned and evaluate it faster. Although that shouldn't be the case b/c a human radiologist should look at every dam thing on that image even if its not mentioned by the computer, i figured id give this guy the benefit of the doubt. At the end of the day if a radiologist can get through more images faster, than there will be less radiologist because there sure as hell won't be more IM or surgery doctors order CTs just because the scanner is open right now

I was only speculating theoretically on how it will progress: computer compares every slice to a bank of standardized normals. Initially, it will only only highlight the problem areas for the radiologist to hone in on (the radiologist will still look at the entire study). This will be beneifical in the case of stat studies and for increasing efficiency. As time progresses, computers would start doing more and more.

I believe they have a similar system for finding calcs on mammograms and finding pulmonary nodules on CXRs. It isn't that great currently. I think computer aided diagnosis will progress in certain areas faster than others... i.e. pulm nodules and finding calcs (renal, breast, etc). I agree; it won't be for some time- likely not in our life time.

In the future, unless something drastically changes, there is only going to be a huge increase in imaging studies and demand for radiology services. America is addicted to imaging... and rightfully so. The objective diagnosis of disease is the next logical step. The endpoint being- a machine scans your body and tells you what wrong with you.

ER/Family/IM docs especially can barely make a serious diagnosis without imaging. The subjective physical exam is not being as heavily emphasized to this generation of clinicians. On top of this, over the next few years, NPs and PAs (physician extenders) are expected to increase "faster than the national average". They will be even more reliant on imaging.

I think the demand for imaging services will continue to expand, but at the same time there will be increased emphasis on evidence based medicine and trying to decrease the reliance on imaging because WAY too many studies are ordered now. There will also be cuts to reimbursement to the older technologies. The question that remains is how it will all balance out.

 

[muncher]

I would be more concerned about the photo recognition technology that is similar to what is used in the Amazon iphone app that allows the user to take a picture of something and then it searches amazon for something similar. From my experience, this works pretty well for common items. I can imagine someone creating a huge data base of normals for say every slice of a body CT (or organ) or head CT/MRI and then using the computer to compare each of the patient's slices to the normal database. The computer would then mark areas of abnormality and draw the radiologist's attention to these areas.

Eventually, you could have a database of abnormals specific to the area of where it finds the abnormality. For example- abnormality in the RLQ in area of appendix compares the patient's images to 1000 varieties/complications of appendicitis, cancer, backwash ileitis, etc... Abnormality in the pancreas- compares with database on pancreatitis, cancers, etc. Clinician suspects stone? The new renal protocol searches the area of the kidneys and ureters for a high density stone. Eventually, the computer could use key patient information (symptoms, age, vitals, labs) to narrow the differential and help focus its search. Of course, I think it would have a lot of problems for someone who has really messed up anatomy...i.e. whipple procedure or someone with mass trauma, but I think specific databases could be established for the different scenarios.

Even with something similar to this, I think there will be an intermediate phase where a radiologist is still absolutely essential. It is really hard to imagine a computer replacing us in our life time.

Don't know if someone commented on this already, but this is actually done by humans =)

 

[ashkenazi1234]

Bump

why so many unfilled radiology spots (41) this year? End of radiology?

 

[LuckyST]

You have to think of it as a percentage. Radiology had 41 unfilled spots out of 1124 offered, which is 3.6% unfilled. The numbers for RadOnc are 6 unfilled out of 171 offered (3.5%), for anesthesia are 51 unfilled out of 1404 offered (3.6%), etc.

So, no, this is not the end of Radiology.

 

[Xerxes1729]

Jeopardy, chess, and other games are all systems with well-defined goals and rules. We create such games to challenge ourselves, and even though they're actually rather simple, we're impressed when computers can perform well in them. There are myriad other tasks that we consider entirely mundane, such as walking down a busy street, which in reality turn out to be extraordinarily hard to get a computer to do. Our perspective on what constitute easy and difficult problems is heavily skewed by the fact that our brains evolved to perform certain tasks extremely well, with little or no conscious effort.

 

[Long Way to Go]

to quote an attending, the worst medical student can read EKG's better than the computer.

Atul Gawande's 1998 article in The New Yorker (later republished in Complications) disagrees. The full study can be read here.

Now, I certainly am not an physician or even a medical student. But I just wanted to bring it up--do you see any flaws with it or know of any other more recent research in the area?

 

[youngdoc8205]

Atul Gawande is commonly known among doctors as someone who rhymes with the word "fooshbag"; the only people I know who buy into the crap he says are the MGH faculty, the media, and the pre-medical student.

I

 

[bronx43]

Jeopardy, chess, and other games are all systems with well-defined goals and rules. We create such games to challenge ourselves, and even though they're actually rather simple, we're impressed when computers can perform well in them. There are myriad other tasks that we consider entirely mundane, such as walking down a busy street, which in reality turn out to be extraordinarily hard to get a computer to do. Our perspective on what constitute easy and difficult problems is heavily skewed by the fact that our brains evolved to perform certain tasks extremely well, with little or no conscious effort.

Anything with an objectively measurably outcome can be better performed by a machine, due to the fact that machines don't make mistakes and can be perpetually improved by use of better starting materials, more efficient mechanics, and more robust algorithms. The only value-add of the human is the ability to innovate - either improve upon or create entirely new structures or ideas. Creating a robot that walks down the street as quickly (I'm inserting this as the objective variable) as humans might be a challenge today, but this technology will surely be available some time in the future if the effort was made.

 

[bronx43]

Atul Gawande is commonly known among doctors as someone who rhymes with the word "fooshbag"; the only people I know who buy into the crap he says are the MGH faculty, the media, and the pre-medical student.

I

Let's try to resist the ad hominems. He might very well be a "fooshbag," but your statement isn't exactly a response to his argument, which is based upon the questioning of "intuitive medicine" and supported by a study published by AHA. He does have his character flaws, but I personally think he brings up a salient point with that particular article.

 

[colbgw02]

Anything with an objectively measurably outcome can be better performed by a machine

This is sort of the whole point though, right? With rare exceptions, image interpretation isn't an "objectively measurably [sic] outcome". I don't know enough about computers to say what will or will not happen, but understanding what I do about the art and nuance of radiology, I sleep soundly at night without fear of being replaced by a machine anytime soon.

 

[hartbot]

Anything with an objectively measurably outcome can be better performed by a machine, due to the fact that machines don't make mistakes and can be perpetually improved by use of better starting materials, more efficient mechanics, and more robust algorithms. The only value-add of the human is the ability to innovate - either improve upon or create entirely new structures or ideas. Creating a robot that walks down the street as quickly (I'm inserting this as the objective variable) as humans might be a challenge today, but this technology will surely be available some time in the future if the effort was made.

But this observation does not mean the end of radiology due to machines. As has been brought up earlier, it may be much more possible for an NP to act as a cardiologist by typing symptoms into the computer and having it spit out some of the most likely causes.

The thing to worry about is how long will it take for machines to become as good with pattern recognition of very complex images that they will be on par with radiologists.

Computers have mastered bar codes, and QR codes, but I also heard that the most intelligent robot cannot tell the difference between a cat or a dog...so I don't know what to think.

 

[MossPoh]

I still think that if such technology WERE to come out that made reading image studies obsolete, it wouldn't make radiology obsolete. It would merely be a shifting set of responsibilities and skills within the field. Nothing is perfect and there will always be mistakes. Even if they were rare, mistakes happen. There'd still need to be checkpoints. There would still need to be someone with some medical knowledge and technical expertise to bridge the gap. People like working with other people. We would also, presumably, be growing and evolving with the technology. I highly doubt there would be a sudden blindside where a radiologist shows up to work and sees his station is gone with a letter saying the computer will handle things for now on. One adapts. If the point comes to where radiologists are replaced by computers, then there will already be lots of other people out of work in pretty much every field out there before hand.

 

[Radonkulous]

...

 

[womp]

Better imaging technology in the future may make radiology obsolete, as it will take less guesswork and expert human input to read.

Thus the future could be a two-hit punch: 1. better imaging tech and 2. better pattern recognition software.

 

[Radonkulous]

...

 

[bronx43]

Interesting. My money is on IR being the next hot thing in medicine, if they are able to control the patient flow. That's a big if at this point.

They wouldn't be able to control patient flow, and therein lies the problem with these support specialties. You can't just go around taking the high paying procedures and call yourself a procedural specialist without the primary field retaliating and doing the procedures and imaging themselves. And let's face it. It's not that hard for the primary specialties to build the training of these procedures into its curriculum. Look at cardiology, and how they were able to take all the procedures from the IR guys, simply because they had the intrinsic advantage of having direct patient influence.

But even if you discount this issue, this isn't the early 1990s, where the budget was balanced and the US wasn't in a quandary economically and financially. It's pretty much inevitable at this point that we'll see a drastic downsizing of the health care sector, and medical care will trend downmarket to primary care and other cost effective measures. That's the main reason I'm rather bearish on these high-end care specialties and bullish on primary care in general.

 

[badasshairday]

They wouldn't be able to control patient flow, and therein lies the problem with these support specialties. You can't just go around taking the high paying procedures and call yourself a procedural specialist without the primary field retaliating and doing the procedures and imaging themselves. And let's face it. It's not that hard for the primary specialties to build the training of these procedures into its curriculum. Look at cardiology, and how they were able to take all the procedures from the IR guys, simply because they had the intrinsic advantage of having direct patient influence.

IR isn't "taking" any procedures... they are creating them.

Also, IR is becoming very clinical. IR's done with training in the past 5 years are MUCH more clinical than those who finished training just a decade ago. At my institution IR has their own clinic, admits their own patients, does all vascular interventions.

 

[PistolPete]

Yeah, it will be interesting to see how IR does in the near future. I really think that IR should branch away from DR, since it really is a separate beast and the people in the field are much different than those in DR.