Jump to content
Premed 101 Forums
Sign in to follow this  
medicine2019

Radiology as a backup...?

Recommended Posts

What would be your first choice? From the rest of your question, it seems that you are really asking whether radiology is decreased in competitiveness enough to use as a backup, but to answer the question (as worded) of whether it's a good idea, you'd have to consider whether you could make your application competitive enough for both fields, and weigh the risk that you appear uncommitted to both.

Share this post


Link to post
Share on other sites

If you had a genuine/potential interest in rads....I'd encourage it!

Years of low competition. Not much reason for it to change. Top paying specialty. Opportunity/demand for clinical work if you're tired of staring at screens. 

Share this post


Link to post
Share on other sites

Why is the competition low? 

2 hours ago, PhD2MD said:

If you had a genuine/potential interest in rads....I'd encourage it!

Years of low competition. Not much reason for it to change. Top paying specialty. Opportunity/demand for clinical work if you're tired of staring at screens. 

 

Share this post


Link to post
Share on other sites
1 hour ago, heydere said:

Why is the competition low? 

 

I think a lot of people see the writing on the wall for rads, AI will probably reduce the demand for rads in the future. 

Share this post


Link to post
Share on other sites
1 hour ago, heydere said:

Why is the competition low? 

 

 

7 minutes ago, Edict said:

I think a lot of people see the writing on the wall for rads, AI will probably reduce the demand for rads in the future. 

I think its a little over-sold, especially now. @Edict, a lot of your defense of the future of anesthesia applies to radiology (obviously not all of it!), plus radiology seems to have a better safety net because they were wise enough to take on most of the interventional procedures. This has resulted in increased demand and plenty to fall back on when AI eventually reads images better (+ the buffer of the CanEHdian health care system accepting the technology).

Happy new years :)

Share this post


Link to post
Share on other sites
19 minutes ago, PhD2MD said:

 

I think its a little over-sold, especially now. @Edict, a lot of your defense of the future of anesthesia applies to radiology (obviously not all of it!), plus radiology seems to have a better safety net because they were wise enough to take on most of the interventional procedures. This has resulted in increased demand and plenty to fall back on when AI eventually reads images better (+ the buffer of the CanEHdian health care system accepting the technology).

Happy new years :)

The thing with rads though is yes the interventional side will be fine, but that is still only a part of the field. Diagnostic radiology probably has less of a defense against AI than anesthesia.  A lot of radiology images are storable, consistent and more black and white than decisions made by anesthesia in the OR. I could see in 15-20 years a scan that takes 30 minutes to read today taking 5 minutes, and that could reduce the number of radiologists needed drastically. Also, a new AI tech would probably be much easier to distribute than an anesthesia robot, since it would just be software that needs to be downloaded and this could reduce the time it takes for the tech to spread. The writing is more on the wall for DR than it is for anesthesia, i could see someone going into anesthesia having no issues for their career, but i can't say the same for radiology. 

 

In the short run i.e the next 10-15 years, i don't see any issues, even if we had the tech today it would take 5-10 years to prove it, and then realistically 5-10 years to make it to the market for rads specifically and this would be optimistic. So if you are interested in DR right now, you probably will find a job, but i still forsee issues emerging by mid-career/late career and theres a good chance by then you'd have the seniority to survive any reductions to demand, but if you are lukewarm about the specialty in the first place, i can see why you would rather not deal with this potential. 

Share this post


Link to post
Share on other sites

I view tech in radiology as just changing/enhancing the job.  The focus will be more on computing and data management, and less on reads.  The volume of imaging is increasing and so tech will just continue to make it more manageable which ultimately should benefit patients and bring down costs.  I don't see a "doomsday"  scenario for radiologists, but rather continued workload and productivity demands, especially with tech developments.  No question AI can do some very impressive things, but ultimately I don't think it's close to being at the stage where a program can write something up or communicate with other physicians regarding specifics  - that will fall to the radiologist (even if guided by tech).  
 

Share this post


Link to post
Share on other sites
2 hours ago, Edict said:

Diagnostic radiology probably has less of a defense against AI than anesthesia.  A lot of radiology images are storable, consistent and more black and white than decisions made by anesthesia in the OR. 

Also differ with this. By the same reasoning, AI could also equally replace other fields of medicine (input patient symptoms/physical exam/labwork for a diagnosis).

We have had computer-assisted detection in mammography for many years now, and most of the findings flagged by the computer are dismissed upon human review (the technology helps as a second reader for increased sensitivity, but if everything flagged was pursued, this would lead to a huge number of workups and anxiety). 

Contrary to common perception of pattern recognition (appearance x = diagnosis y), much of image interpretation is far from black and white, and it may be necessary to decide whether a finding is even real or could be within the range of normal, using one's knowledge of additional patient factors / history and value system to help decide how to word the final report and recommendations.

The other question of course is who will shoulder responsibility if the machine gets it wrong, so it is likely that technology will continue to serve as an aid to human physicians as tere said.

Share this post


Link to post
Share on other sites
On 1/1/2019 at 11:48 AM, heydere said:

Why is the competition low? 

 

Probably because of videos like these

AI will never be able to replace radiology entirely, but the fear of the uncertainty is probably what drives potential medical students away from this field. It will likely be at least a decade before we see this is implemented in clinical practice. Since most current medical students/residents have a career span of at least 25 years, it is highly likely that they will encounter this in their lifetime.

Edited by ArchEnemy

Share this post


Link to post
Share on other sites
On 1/1/2019 at 3:41 PM, Lactic Folly said:

The other question of course is who will shoulder responsibility if the machine gets it wrong, so it is likely that technology will continue to serve as an aid to human physicians as tere said.

This is a really interesting question. The best explanations I've heard are that once the error rate for AI drops low enough, it becomes more economical for the hospital/government/whoever if providing health care to take that responsibility as a calculated cost, than it would be to employ staff to do the reading.

Share this post


Link to post
Share on other sites
5 hours ago, ArchEnemy said:

Probably because of videos like these

AI will never be able to replace radiology entirely, but the fear of the uncertainty is probably what drives potential medical students away from this field. It will likely be at least a decade before we see this is implemented in clinical practice. Since most current medical students/residents have a career span of at least 25 years, it is highly likely that they will encounter this in their lifetime.

Yes, this

Share this post


Link to post
Share on other sites
On 1/1/2019 at 3:41 PM, Lactic Folly said:

Also differ with this. By the same reasoning, AI could also equally replace other fields of medicine (input patient symptoms/physical exam/labwork for a diagnosis).

We have had computer-assisted detection in mammography for many years now, and most of the findings flagged by the computer are dismissed upon human review (the technology helps as a second reader for increased sensitivity, but if everything flagged was pursued, this would lead to a huge number of workups and anxiety). 

Contrary to common perception of pattern recognition (appearance x = diagnosis y), much of image interpretation is far from black and white, and it may be necessary to decide whether a finding is even real or could be within the range of normal, using one's knowledge of additional patient factors / history and value system to help decide how to word the final report and recommendations.

The other question of course is who will shoulder responsibility if the machine gets it wrong, so it is likely that technology will continue to serve as an aid to human physicians as tere said.

The technology though is advancing at a rapid pace, and it wouldn't be shocking that they will have this down in 10-15 years as a concept. It may take another 10-15 years for that technology to truly land in the hands of radiologists depending on the center (academic centers may begin using it earlier). I don't think these technologies could replace diagnostic rads entirely, but i could definitely see demand being reduced. Time to read scans can go down which would mean fewer radiologists are needed to run a department, this could ultimately mean less jobs, like i mentioned above, people currently in med school and residents today probably have less to be worried about. Once this technology truly lands in the hands of radiologists across Canada, most of the people now will have already been hired and likely will be able to stay employed, they just won't hire new radiologists. Radiology departments are expensive and as governments get tight on cash and demand for imaging keeps going up, this kind of technology will probably be adopted by hospitals and governments around the world. 

 

 

Share this post


Link to post
Share on other sites
On 1/1/2019 at 1:08 PM, PhD2MD said:

 

I think its a little over-sold, especially now. @Edict, a lot of your defense of the future of anesthesia applies to radiology (obviously not all of it!), plus radiology seems to have a better safety net because they were wise enough to take on most of the interventional procedures. This has resulted in increased demand and plenty to fall back on when AI eventually reads images better (+ the buffer of the CanEHdian health care system accepting the technology).

Happy new years :)

well sort of....the issue is in the US for instance interventional radiology is now moving away to its own separate field. There is talk it may do the same thing in Canada. That doesn't mean there still isn't a ton of procedures rads do - 1/3 of the job is procedures in the community often and none of that is really what most people would call interventional rads really. 

 

Share this post


Link to post
Share on other sites
21 hours ago, Edict said:

The technology though is advancing at a rapid pace, and it wouldn't be shocking that they will have this down in 10-15 years as a concept. It may take another 10-15 years for that technology to truly land in the hands of radiologists depending on the center (academic centers may begin using it earlier). 

Ok, this all sounds good in theory. However, say you are at one of the academic hospitals you have trained at. You wish to request a CT or MR on an inpatient. How does this work? A handwritten paper request is faxed to the radiology department. These faxed sheets are collected in a folder. To keep track of the requests and their priority, the staff keep a handwritten list of all the patient names on a sheet of paper. They update the queue as necessary by writing a new list of patient names on a new sheet of paper, by hand. It is currently 2019. What does this imply about diffusion of technology, given how long computers have been in existence?

And as rmorelan said, image reading is only one part of the role. A good portion of the day can be on the phone/email or away from the workstation doing procedures, dealing with tech/nurse/patient concerns, protocolling, interacting with referring physicians/presenting at rounds/teaching as applicable.

Share this post


Link to post
Share on other sites
1 hour ago, Lactic Folly said:

Ok, this all sounds good in theory. However, say you are at one of the academic hospitals you have trained at. You wish to request a CT or MR on an inpatient. How does this work? A handwritten paper request is faxed to the radiology department. These faxed sheets are collected in a folder. To keep track of the requests and their priority, the staff keep a handwritten list of all the patient names on a sheet of paper. They update the queue as necessary by writing a new list of patient names on a new sheet of paper, by hand. It is currently 2019. What does this imply about diffusion of technology, given how long computers have been in existence?

And as rmorelan said, image reading is only one part of the role. A good portion of the day can be on the phone/email or away from the workstation doing procedures, dealing with tech/nurse/patient concerns, protocolling, interacting with referring physicians/presenting at rounds/teaching as applicable.

Yes, its very true that in some things healthcare is unbelievably ancient, and EMR is one of those. Canada though tends to fund things that will make a difference in patient care as well as things that are flashy. For example, they will fund the latest CT/MRI machines, hybrid ORs, surgical toys, but they won't change EMRs, paper based resources that are considerably less exciting and easily explainable to the public. 

Also, because some older physicians prefer to use pen and paper, and these physicians remain in powerful positions, these kinds of old techniques have remained. At the same time, there isn't much political will to try to use tech to get rid of these paper pushers, no one is funding studies to show that an electronic system of CT/MR is better than pen and paper. Also, some centers are now moving towards electronic order entry even for CTs/MRIs. 

However, i believe software reading scans may be different. If research can show that the software is consistently better than radiologists at diagnosis, it becomes a hot topic issue, there becomes political will to make it happen. It is just like how we spend millions on the latest CT scanner, but we will still use pen and paper for EMR. When there is a political and public will, these things get accelerated. As you can see the well publicized competitions between software and humans, there will be this kind of public attention. This is in part because this is a concept that is easily explained and understandable to the public.

Like i mentioned elsewhere, there will be a radiology lobby that will argue against this technology and in many cases they may be right, and so the rollout may be delayed, it may affect certain types of imaging more than others, but if you look at the career of someone who is currently 25 in medical school and plans to retire at 65, that is a 40 year career and it would not shock me that in 30 years we would have this kind of tech on radiology desks around the country. It won't replace all radiologists, and it won't even replace a proportional number of radiologists, but it would reduce the number of radiologists needed to read a certain number of scans. It is likely that low risk imaging will be replaced first over complex, high risk imaging.  

Lastly, yes a radiologist's work isn't only image reading, so i don't think just because we read images 6x faster that we will need 1/6th the number of radiologists. This sort of change will come slowly and gradually, it is very possible that the number of radiologists needed will be halved if all scans were read 6x faster for example and like i mentioned before, it is unlikely they are going to kick radiologists out who are mid career. These things will happen slowly, retiring radiologists won't be replaced, there may be less work for existing radiologists, fewer radiology residents, more opportunities in other related fields, more opportunities to do research etc. There will also be a compensatory increase in imaging ordered if scans become easier and quicker to read, so it won't be a dramatic drop in the number of radiologists immediately. 

 

Share this post


Link to post
Share on other sites
1 hour ago, Edict said:

, yes a radiologist's work isn't only image reading, so i don't think just because we read images 6x faster that we will need 1/6th the number of radiologists. This sort of change will come slowly and gradually, it is very possible that the number of radiologists needed will be halved if all scans were read 6x faster for example and like i mentioned before, it is unlikely they are going to kick radiologists out who are mid career. These things will happen slowly, retiring radiologists won't be replaced, there may be less work for existing radiologists, fewer radiology residents, more opportunities in other related fields, more opportunities to do research etc. There will also be a compensatory increase in imaging ordered if scans become easier and quicker to read, so it won't be a dramatic drop in the number of radiologists immediately. 

 

That's assuming the demand for imaging stays constant.  No reason that isn't going to be an increase in demand with an ageing population along with potentially a lower cost.  Cost from what I understand is a major reason why more imaging isn't done. 

Ultimately someone has to manage the technology and technicians, communicate with physicians, etc..  and so I see radiologists as becoming more hybrid data scientists.  Any kind of IT interfacing also costs money, so in a way if radiologists embraced the informatics side this could also help them provide a valued role.

I also think pathology will undergo a similar transformation (longer down the line because of data collection issues).

Finally, as someone who has studied some of underlying AI algorithms like neural nets and support vector machines, it's worth mentioning that these work well with well-defined problems like classification.  Natural language understanding, or the more general AI problem is much harder - Watson, the jeopardy champion, for all its results on the game show has been limited as medical oncology aid with disappointing results even after a lot of initial publicity.  So in a sense, what AI can/cannot do is somewhat limited which is why radiologists may have a great opportunity to expand their scope and provide greater service, albeit by potentially altering some of their focus (with more data science rather than manual reads).  

 

Share this post


Link to post
Share on other sites
2 hours ago, Edict said:

Yes, its very true that in some things healthcare is unbelievably ancient, and EMR is one of those. Canada though tends to fund things that will make a difference in patient care as well as things that are flashy. For example, they will fund the latest CT/MRI machines, hybrid ORs, surgical toys, but they won't change EMRs, paper based resources that are considerably less exciting and easily explainable to the public. 

Also, because some older physicians prefer to use pen and paper, and these physicians remain in powerful positions, these kinds of old techniques have remained. At the same time, there isn't much political will to try to use tech to get rid of these paper pushers, no one is funding studies to show that an electronic system of CT/MR is better than pen and paper. Also, some centers are now moving towards electronic order entry even for CTs/MRIs. 

However, i believe software reading scans may be different. If research can show that the software is consistently better than radiologists at diagnosis, it becomes a hot topic issue, there becomes political will to make it happen. It is just like how we spend millions on the latest CT scanner, but we will still use pen and paper for EMR. When there is a political and public will, these things get accelerated. As you can see the well publicized competitions between software and humans, there will be this kind of public attention. This is in part because this is a concept that is easily explained and understandable to the public.

Like i mentioned elsewhere, there will be a radiology lobby that will argue against this technology and in many cases they may be right, and so the rollout may be delayed, it may affect certain types of imaging more than others, but if you look at the career of someone who is currently 25 in medical school and plans to retire at 65, that is a 40 year career and it would not shock me that in 30 years we would have this kind of tech on radiology desks around the country. It won't replace all radiologists, and it won't even replace a proportional number of radiologists, but it would reduce the number of radiologists needed to read a certain number of scans. It is likely that low risk imaging will be replaced first over complex, high risk imaging.  

Lastly, yes a radiologist's work isn't only image reading, so i don't think just because we read images 6x faster that we will need 1/6th the number of radiologists. This sort of change will come slowly and gradually, it is very possible that the number of radiologists needed will be halved if all scans were read 6x faster for example and like i mentioned before, it is unlikely they are going to kick radiologists out who are mid career. These things will happen slowly, retiring radiologists won't be replaced, there may be less work for existing radiologists, fewer radiology residents, more opportunities in other related fields, more opportunities to do research etc. There will also be a compensatory increase in imaging ordered if scans become easier and quicker to read, so it won't be a dramatic drop in the number of radiologists immediately. 

 

I agree with a lot of this. Autopilot didn't replace pilots, but there has been a substantial decrease in the prestige and pay of pilots as technology has streamlined the airline industry. 

Share this post


Link to post
Share on other sites
7 hours ago, blah1234 said:

I agree with a lot of this. Autopilot didn't replace pilots, but there has been a substantial decrease in the prestige and pay of pilots as technology has streamlined the airline industry. 

There is the tech aspect. But airline travel.used to be for only the rich due to cost, so everything associated with it was glamorous in the public eye. Now that airline travel is so cheap and available to the masses, it's essentially similar to bus travel in the public's eye, so the pilot no longer has that aura about him/her. 

Share this post


Link to post
Share on other sites
6 hours ago, NLengr said:

There is the tech aspect. But airline travel.used to be for only the rich due to cost, so everything associated with it was glamorous in the public eye. Now that airline travel is so cheap and available to the masses, it's essentially similar to bus travel in the public's eye, so the pilot no longer has that aura about him/her. 

That's a great point that I didn't think of. I can see medicine going down that route as well in terms of public perception. I have had multiple mentors tell me how much has changed in terms of patient respect and the public perception of doctors. 

Share this post


Link to post
Share on other sites
12 hours ago, blah1234 said:

That's a great point that I didn't think of. I can see medicine going down that route as well in terms of public perception. I have had multiple mentors tell me how much has changed in terms of patient respect and the public perception of doctors. 

Its already been happening with the rise of "Dr. Google" and patients emphatically informing you that they "did their research" on intervention "x, y, or z" in the primary care setting (especially), but also to a lesser extent in the specialist clinic.

Shared decision making is great. Greater access to information (read: the internet) is great. Patients educating themselves on their condition(s) and treatment options is great.

Engaging with patients reading unscientific internet sources and reaching inflexible conclusions on treatment options, or causing themselves unnecessary anxiety through googling symptoms for which they do not possess the training to differentiate or put into the proper context is...less than satisfying.

Anecdotally, you definitely notice a difference in how patients interact with their primary care provider as compared to the medical or surgical specialist for who they have often had to wait a considerable amount of time to see or whose services they have had to advocate for.

Share this post


Link to post
Share on other sites
On 1/3/2019 at 10:44 PM, tere said:

That's assuming the demand for imaging stays constant.  No reason that isn't going to be an increase in demand with an ageing population along with potentially a lower cost.  Cost from what I understand is a major reason why more imaging isn't done. 

Ultimately someone has to manage the technology and technicians, communicate with physicians, etc..  and so I see radiologists as becoming more hybrid data scientists.  Any kind of IT interfacing also costs money, so in a way if radiologists embraced the informatics side this could also help them provide a valued role.

I also think pathology will undergo a similar transformation (longer down the line because of data collection issues).

Finally, as someone who has studied some of underlying AI algorithms like neural nets and support vector machines, it's worth mentioning that these work well with well-defined problems like classification.  Natural language understanding, or the more general AI problem is much harder - Watson, the jeopardy champion, for all its results on the game show has been limited as medical oncology aid with disappointing results even after a lot of initial publicity.  So in a sense, what AI can/cannot do is somewhat limited which is why radiologists may have a great opportunity to expand their scope and provide greater service, albeit by potentially altering some of their focus (with more data science rather than manual reads).  

  

My last point was basically your first point, yes imaging demand will go up which will soften the blow. I think however that like others have rightly mentioned, radiology has been a rising star so far, record levels of compensation, record demand, but if AI begins to read imaging there will be drops in pay and demand. Pilots used to be demigods, "your life in their hands", but with the safety systems in place, pilots have become monitors, backups, ready to step in. Demand hasn't dropped but income and prestige have.

There will always be a role for radiology, but it just may not be the same as what we know today. 

Share this post


Link to post
Share on other sites
2 hours ago, Edict said:

My last point was basically your first point, yes imaging demand will go up which will soften the blow. I think however that like others have rightly mentioned, radiology has been a rising star so far, record levels of compensation, record demand, but if AI begins to read imaging there will be drops in pay and demand. Pilots used to be demigods, "your life in their hands", but with the safety systems in place, pilots have become monitors, backups, ready to step in. Demand hasn't dropped but income and prestige have.

There will always be a role for radiology, but it just may not be the same as what we know today. 

I've also suggested from the beginning that radiology will change!  I think it's too early to tell whether the change will be more similar to pilots or whether radiologists will find ways to actively incorporate the tech for greater proficiency and analysis.  Many radiologists seem to be aware and engaged with the changes and seem to be looking at ways to improve their workflow.  I'm glad you see the greater imaging demand side too - your initial posts seemed a lot more pessimistic regarding workforce. 

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
Sign in to follow this  

×
×
  • Create New...