MD Radiology vs Internal Medicine

[cecumbowels]

I am currently interested in both radiology and internal medicine. I really like radiology in that imaging is really interesting to me, technology oriented field, possibility of working from anywhere, and helping a large number of patients in one single day. I would apply to diagnostic radiology and then the goal would be interventional rad. I also like internal medicine though. I like that you have a broad knowledge, variety of pathology, possibility of subspecialization and keeping options open, unlike most people I Iike talking to patients. Things I dont like about IM include that you have to deal with a lot of stuff and everyone tends to dump stuff on IM. I think lifestyles of both specialties are similar. My step 1 was a 215 and step 2 of 230. Any advice of which one might be a better option or fit?

---

Members don't see this ad.

 

[differentiating]

I am currently interested in both radiology and internal medicine. I really like radiology in that imaging is really interesting to me, technology oriented field, possibility of working from anywhere, and helping a large number of patients in one single day. I would apply to diagnostic radiology and then the goal would be interventional rad. I also like internal medicine though. I like that you have a broad knowledge, variety of pathology, possibility of subspecialization and keeping options open, unlike most people I Iike talking to patients. Things I dont like about IM include that you have to deal with a lot of stuff and everyone tends to dump stuff on IM. I think lifestyles of both specialties are similar. My step 1 was a 215 and step 2 of 230. Any advice of which one might be a better option or fit?

Have you done any rads (dr or ir) rotations? How did they go?

 

[radsisrad]

Radiology is a great field, but its future is uncertain - whether because of corporate take-over (and consequent bitter deals for rads) or AI. If you can handle IM day-to-day practice, I'd recommend IM residency followed by a lucrative specialization (cardiology or GI, in particular). Especially seeing as you seem more inclined toward procedures (e.g., IR).

 

[cecumbowels]

I definitely like doing procedures. Do you think I would even have a chance at those kinds of subspecialties because my step 1 is 215 and step 2 230? I hear those subspecialties are pretty competitive.

 

[Dave1980]

Radiology is a great field, but its future is uncertain - whether because of corporate take-over (and consequent bitter deals for rads) or AI. If you can handle IM day-to-day practice, I'd recommend IM residency followed by a lucrative specialization (cardiology or GI, in particular). Especially seeing as you seem more inclined toward procedures (e.g., IR).

The future of the field is NOT uncertain. Don't listen to chicken little.

IR fellowship is now 2 years long so you should consider going straight into IR if that is your plan.

 

[cecumbowels]

My step 1 is 215. I dont think I would be able to go straight into IR.

 

[radsisrad]

I definitely like doing procedures. Do you think I would even have a chance at those kinds of subspecialties because my step 1 is 215 and step 2 230? I hear those subspecialties are pretty competitive.

I doubt that your Step scores will matter much when it comes to fellowship apps (unless you're applying to prestigious, "top-tier" ones... and there's no need to do so). Perform well in residency and get good LORs, and you'll be fine, I'm sure.

The future of the field is NOT uncertain. Don't listen to chicken little.

IR fellowship is now 2 years long so you should consider going straight into IR if that is your plan.

OP can gauge who's being hyperbolic. Did Chicken Little not proclaim that "The sky is falling!"? I've said no such thing. The frank reality is that the future of radiology - specifically, DR - is uncertain. Few deny this. Now, we can debate whether AI will render radiologists utterly obsolete (unlikely) or lead to a substantial oversupply of DRs (more likely over the coming decades), whether rads will continue to enjoy the income we still do given the changing healthcare landscape, etc. But regardless of which specialty one pursues, recognize that very few specialties enjoy near-certainty regarding the future.

 

[Dave1980]

The frank reality is that the future of radiology - specifically, DR - is uncertain. Few deny this.

Actually almost everyone who knows anything about Radiology denies this. I am a Radiologist and was at RSNA where AI was featured prominently and no one who seriously understands the issue thinks AI is going to replace radiologists anytime even remotely soon.

At the risk of sounding like a total jerk I do think that you as a pre-health student should not be giving advice to medical students about which specialty to go into.

 

[radsisrad]

Actually almost everyone who knows anything about Radiology denies this. I am a Radiologist and was at RSNA where AI was featured prominently and no one who seriously understands the issue thinks AI is going to replace radiologists anytime even remotely soon.

At the risk of sounding like a total jerk I do think that you as a pre-health student should not be giving advice to medical students about which specialty to go into.

I'm a rads resident.

And I'm not stating that AI is going to replace radiologists even remotely soon*. What I'm saying is that the matter is uncertain. And that's probably the concern of least importance as things currently stand. VC/corporate radiology is a spreading phenomenon, and more and more attending rads are reporting sharp dissatisfaction with the field's landscape.

The bottom line is that the safest fields are those in which the physicians "own" their patients - ideally with some procedural components.

*though one must ask oneself if this is really the field to pursue at the age of 26 or so, when most med students graduate

 

[7331poas]

Actually almost everyone who knows anything about Radiology denies this. I am a Radiologist and was at RSNA where AI was featured prominently and no one who seriously understands the issue thinks AI is going to replace radiologists anytime even remotely soon.

At the risk of sounding like a total jerk I do think that you as a pre-health student should not be giving advice to medical students about which specialty to go into.

There is no doubt in my mind that AI will eventually replace all human labor. Whether than is "remotely soon" is hard to say. But certainly DR and path are the most susceptible.

 

[Gurby]

Actually almost everyone who knows anything about Radiology denies this. I am a Radiologist and was at RSNA where AI was featured prominently and no one who seriously understands the issue thinks AI is going to replace radiologists anytime even remotely soon.

At the risk of sounding like a total jerk I do think that you as a pre-health student should not be giving advice to medical students about which specialty to go into.

At least one attending thinks the future is trending more towards the "apocalyptic" end of the spectrum, and managed to get his opinion published in JACR.

https://www.jacr.org/article/S1546-1440(18)30389-2/fulltext

Also the issue isn't so much "replacing" as it is increasing efficiency to the point that fewer radiologists are needed and the job market tanks. We still pay cardiologists to look at EKG's so I don't imagine we'll trust AI to fly solo any time soon.

 

[LebronManning]

Internal Medicine. Way more flexibility

 

[Splenda88]

My step 1 is 215. I dont think I would be able to go straight into IR.

Your score might be an issue for DR, but you never know. Someone in my class matched DR with the same step 1 score but that was in 2018. As an IM, I would vote for DR unless you plan to do a fellowship.

 

[Angus Avagadro]

Your score might be an issue for DR, but you never know. Someone in my class matched DR with the same step 1 score but that was in 2018. As an IM, I would vote for DR unless you plan to do a fellowship.

I believe DR was more competitive this year, so 215 might be a little light. There are some very good community residencies that might be interested in you. Do your due diligence on the program. Good luck and best wishes.

 

[Womb Raider]

215 is very low for DR but not impossible. In the 2018 Charting Outcomes 46/59 applicants with STEP 1 between 211-220 matched, but it was also ~7% of the total applicant pool with 93% having higher step scores. If you like IR, a good number of programs have ESIR now so your path to IR would only take 6 years. I would strongly encourage you to dual apply though since you're on the weaker side for DR. Start building relationships with faculty for both specialties so you can have solid letters in both.

 

[7331poas]

215 is very low for DR but not impossible. In the 2018 Charting Outcomes 46/59 applicants with STEP 1 between 211-220 matched, but it was also ~7% of the total applicant pool with 93% having higher step scores. If you like IR, a good number of programs have ESIR now so your path to IR would only take 6 years. I would strongly encourage you to dual apply though since you're on the weaker side for DR. Start building relationships with faculty for both specialties so you can have solid letters in both.

I thoughtESIR is going away. Also even if there are ESIR spots, its not certain you will get those spots over your fellow residents.

 

[deleted746658]

Radiology is a great field, but its future is uncertain - whether because of corporate take-over (and consequent bitter deals for rads) or AI. If you can handle IM day-to-day practice, I'd recommend IM residency followed by a lucrative specialization (cardiology or GI, in particular). Especially seeing as you seem more inclined toward procedures (e.g., IR).

as someone also entering radiology, i would have to agree with the above. if you could see yourself happy in either, i would pursue IM.

DR is a sweet field, and i chose it despite being fully aware of the aforementioned uncertainties, because i couldn't see myself happy in any other field. that said, i'm not super confident i'll be able to finish my career as a diagnostic radiologist. AI/ML will undoubtedly impact the field and change how DR is practiced. how DR evolves - imo - will largely be determined by who controls these new technologies. if VC and corps are in control, buh-bye. if DR's can integrate into clinical practice and be the first clinician's to successfully work in symbiosis with AI, we might be alright. tho, i do believe it will definitely decrease the # of radiologists needed in the workforce, and if the governing bodies are not proactive about this (start reducing # of trainees in the next 3-7 years), we could find ourselves with a terribly oversaturated job market

i also find it concerning that many of the older rad docs i've spoken with don't seem to be alarmed by AI. they often cite where the tech is at today, in the present moment, and fail to speculate on how swiftly it will progress in the next 5-15 years... some of the biggest players (google, facebook, microsoft) are spending billions with the brightest minds working on these projects. let that sink in. maybe if i had a nice nest egg i wouldn't be too concerned either, but the reality for the new generation of docs with our large debt burdens, a disruption as predicted by AI experts, would be (financially, professionally) devastating...

 

[Womb Raider]

I thoughtESIR is going away. Also even if there are ESIR spots, its not certain you will get those spots over your fellow residents.

ESIR is brand new, why would it be going away? You’ve been misinformed. From my time on the interview trail the general consensus was that getting ESIR spots was not difficult since most people end up disliking the lifestyle associated with IR. Only time will tell since it’s so new but I think it’s a very viable option currently.

 

[radsisrad]

IR is not a viable DR-bailout strategy in the case of AI disruption. There are only so many PICC, PAD, and cancer cases to go around. Besides, if (when?) AI does become startlingly useful and independent in bread-and-butter rads functions, there will be a frantic stampede of residents and attendings to IR fellowship spots. What's the saying? "May the odds be ever in your favor."

Will DR be significantly damaged by AI? I hope not. But don't take the word of someone who promises you that it won't.

 

[Gurby]

IR is not a viable DR-bailout strategy in the case of AI disruption. There are only so many PICC, PAD, and cancer cases to go around. Besides, if (when?) AI does become startlingly useful and independent in bread-and-butter rads functions, there will be a frantic stampede of residents and attendings to IR fellowship spots. What's the saying? "May the odds be ever in your favor."

Will DR be significantly damaged by AI? I hope not. But don't take the word of someone who promises you that it won't.

I wonder if DR may be more resilient in some ways - as you point out "bread and butter rads functions" are the primary targets of AI development. If IR's in private practice are spending 50% of their time doing procedures and 50% of their time reading bread and butter studies, they might stand to lose more than a fellowship-trained DR who can still bank on the fringe cases and more specialized studies that will be much harder to develop AI for (and that reimburse better anyways).

 

[Jyggaswoop]

any possiblity that salaries can be like fam med in the sense that family med docs utilize NP's to see more patients and make more money. AI can help speed read the bread and butter things, and hospitals will still want a radiologist to "okay" these in case a law suit comes forth etc. so you get reimbursed less due to AI but you 'okay' a lot more readings.

 

[Angus Avagadro]

With everyone clamoring for single payer, Docs wont be allowed to order imaging studies as dictated by the protocols set by the govt. I was in Scotland 2 yrs ago and front page story was about 20 month girl with intractable nausea and complaints of burning lips. 3 trips to GP who diagnosed flu, then on last trip father was biting her fingers to keep her awake. Later had massive UGI bleed. Turns out she swallowed a circular battery. I have met many UK trained docs, and they are very well trained. They just have their hands tied by the National Health Service. A $15 xray might have saved the day.

 

[nimbus]

any possiblity that salaries can be like fam med in the sense that family med docs utilize NP's to see more patients and make more money. AI can help speed read the bread and butter things, and hospitals will still want a radiologist to "okay" these in case a law suit comes forth etc. so you get reimbursed less due to AI but you 'okay' a lot more readings.

Yes AI will be the NP’s of radiology.

 

[epimed22]

I just finished my 1st year of medical school, and now I'm doing research at UAB. I will have the chance to have a first author publication and a 2nd publication as a co-author. The project is more related to community health/preventive medicine (I have an MPH). Would this help me if I want to get into a good IM program ???? (thinking that I will get 230+ on step 1)

 

[Lnsean]

Actually almost everyone who knows anything about Radiology denies this. I am a Radiologist and was at RSNA where AI was featured prominently and no one who seriously understands the issue thinks AI is going to replace radiologists anytime even remotely soon.

At the risk of sounding like a total jerk I do think that you as a pre-health student should not be giving advice to medical students about which specialty to go into.

Why does AI have to replace radiologists for it to bad? It can get pretty bad before that. All AI needs to do is make it more efficient to read and interpret studies, and the market for new grads will take a major dump. Playing devils advocate, I think those that are too deep in the field also have their heads buried to things happening around them.

 

[Oh_Gee]

as someone also entering radiology, i would have to agree with the above. if you could see yourself happy in either, i would pursue IM.

DR is a sweet field, and i chose it despite being fully aware of the aforementioned uncertainties, because i couldn't see myself happy in any other field. that said, i'm not super confident i'll be able to finish my career as a diagnostic radiologist. AI/ML will undoubtedly impact the field and change how DR is practiced. how DR evolves - imo - will largely be determined by who controls these new technologies. if VC and corps are in control, buh-bye. if DR's can integrate into clinical practice and be the first clinician's to successfully work in symbiosis with AI, we might be alright. tho, i do believe it will definitely decrease the # of radiologists needed in the workforce, and if the governing bodies are not proactive about this (start reducing # of trainees in the next 3-7 years), we could find ourselves with a terribly oversaturated job market

i also find it concerning that many of the older rad docs i've spoken with don't seem to be alarmed by AI. they often cite where the tech is at today, in the present moment, and fail to speculate on how swiftly it will progress in the next 5-15 years... some of the biggest players (google, facebook, microsoft) are spending billions with the brightest minds working on these projects. let that sink in. maybe if i had a nice nest egg i wouldn't be too concerned either, but the reality for the new generation of docs with our large debt burdens, a disruption as predicted by AI experts, would be (financially, professionally) devastating...

mine said he didn't really care about AI because he'd probably be retired by the time AI had any impact lol

 

[Gurby]

Why does AI have to replace radiologists for it to bad? It can get pretty bad before that. All AI needs to do is make it more efficient to read and interpret studies, and the market for new grads will take a major dump. Playing devils advocate, I think those that are too deep in the field also have their heads buried to things happening around them.

At the same time, Radiology more than any other specialty is positioning itself to coexist with the new AI overlords (who are almost certainly going to bring major changes to every field of medicine within the span of our careers). AI is a major topic of interest at conferences, and there is even a subjournal of Radiology devoted specifically to AI. I think it's a bit unfair to say that people have their heads buried (though even new attendings don't feel the heat quite the same way as we students do, being 8 years away from seeing a paycheck...).

Just like the invention of the sewing machine didn't put tailors out of business but actually increased demand for them... I think there is a good chance that as we find more applications for AI in diagnostics, we'll find new niches for radiologists to fill. Or not, who knows.

 

[Lnsean]

At the same time, Radiology more than any other specialty is positioning itself to coexist with the new AI overlords (who are almost certainly going to bring major changes to every field of medicine within the span of our careers). AI is a major topic of interest at conferences, and there is even a subjournal of Radiology devoted specifically to AI. I think it's a bit unfair to say that people have their heads buried (though even new attendings don't feel the heat quite the same way as we students do, being 8 years away from seeing a paycheck...).

Just like the invention of the sewing machine didn't put tailors out of business but actually increased demand for them... I think there is a good chance that as we find more applications for AI in diagnostics, we'll find new niches for radiologists to fill. Or not, who knows.

Oh for sure the corporate bean counters will jump all over this in the name of "patient outcomes" and such. Also, how many tailors do you know personally? The point of sewing machines isn't that they put every tailor out of business. Sewing machines led to mass production (increased scale and efficient) and less tailors. One tailor can make 100 shirts in a day now on a sewing machine that was once done by hand and maybe took 20 tailors to do. Obviously technology is inevitable and cost will have to come down; I'm just saying that people should not be so oblivious to what's happening here: there's a common theme.

 

[Gurby]

Oh for sure the corporate bean counters will jump all over this in the name of "patient outcomes" and such. Also, how many tailors do you know personally? The point of sewing machines isn't that they put every tailor out of business. Sewing machines led to mass production (increased scale and efficient) and less tailors. One tailor can make 100 shirts in a day now on a sewing machine that was once done by hand and maybe took 20 tailors to do. Obviously technology is inevitable and cost will have to come down; I'm just saying that people should not be so oblivious to what's happening here: there's a common theme.

In the short and medium term, sewing machines increased the need for tailors/seamstresses by opening up new markets and increasing demand for goods. Clothes got more fancy and required more stitching to make, everyone owned more clothes, and people got more stuff adjusted to fit them. It took a good 150 years before we started outsourcing all of that work to Chinese sweat shops...

The long term outlook is that everybody gets replaced by AI. The short term outlook is business as usual. The timeline of what happens in between could go either way, I think...

Maybe AI makes radiologists more efficient and tanks the job market. Or maybe since AI is so good at answering yes/no questions and detecting cancers on low dose scans, screening guidelines change and we start scanning everybody all the time for everything (even moreso than we do now). This drives up the number of studies needing review, number of fringe cases needing human eyes, number of more specialized studies ordered that don't have AI support to investigate things discovered by all the scans, etc. I think there's a good chance of there being a golden period of rads becoming even more essential and in-demand thanks to AI, at least in the short-to-medium term.

 

[messengers]

If the OP doesn't mind the intrusion would someone mind going into radiology vs pathology?

I can start another thread if it is too off topic.

 

[Dro133]

Or maybe since AI is so good at answering yes/no questions and detecting cancers on low dose scans, screening guidelines change and we start scanning everybody all the time for everything (even moreso than we do now).

I don't quite understand this point. If there was an overall mortality benefit to be gained from pan-scanning everyone, wouldn't we already be doing so? I'm not sure how AI changes that.

 

[radsisrad]

If the OP doesn't mind the intrusion would someone mind going into radiology vs pathology?

I can start another thread if it is too off topic.

Path is a whole other level of non-social practice. Not as remunerative as rads, but a lot less stressful. Check out their forum to see what the market's like... it was abysmal as of a few years ago, but things may have changed.

 

[Gurby]

I don't quite understand this point. If there was an overall mortality benefit to be gained from pan-scanning everyone, wouldn't we already be doing so? I'm not sure how AI changes that.

Maybe I'm way off the mark here. My thinking is that while we're still far away from "Watson, interpret this CT scan", AI is rapidly getting better than humans at answering questions like "Is there X present?" Is there a pneumothorax, is there a head bleed, is there a lung nodule? It's hard to predict what the downstream implications of this will be as AI continues to improve.

My understanding is that with the current state of medicine, basically the reason we don't pan-scan people (cost aside) is that we find too many incidentalomas that need to be investigated.

I think it's possible that due to the ability of AI to consider so many factors that the human eye cannot, we could see major paradigm shifts in screening and the ability to tell benign vs malignant based on imaging alone. There's an entire new area of study called radiomics, which has only existed for <7 years, dedicated to interpreting the nearly limitless number of "imaging biomarkers" and statistical relationships that can be extracted from imaging.

People apparently thought CT was going to make radiologists obsolete because "it's so easy, everyone will just read their own scans" - look how that turned out. I think that as AI gets better and better, the amount of information available is just going to exponentially increase. Clinicians will want and need help with interpreting that data and figuring out how best to apply it for each individual patient.

Or maybe not. Who knows

 

[Lnsean]

In the short and medium term, sewing machines increased the need for tailors/seamstresses by opening up new markets and increasing demand for goods. Clothes got more fancy and required more stitching to make, everyone owned more clothes, and people got more stuff adjusted to fit them. It took a good 150 years before we started outsourcing all of that work to Chinese sweat shops...

The long term outlook is that everybody gets replaced by AI. The short term outlook is business as usual. The timeline of what happens in between could go either way, I think...

Maybe AI makes radiologists more efficient and tanks the job market. Or maybe since AI is so good at answering yes/no questions and detecting cancers on low dose scans, screening guidelines change and we start scanning everybody all the time for everything (even moreso than we do now). This drives up the number of studies needing review, number of fringe cases needing human eyes, number of more specialized studies ordered that don't have AI support to investigate things discovered by all the scans, etc. I think there's a good chance of there being a golden period of rads becoming even more essential and in-demand thanks to AI, at least in the short-to-medium term.

This is saying nothing really. Obviously technology will advance and we will have to adapt; that was never a point of contention or confusion. It's a part of human history. I don't even know why you even typed all that when it could just be said in a few words.

My point is in the process of adapting, there will be a generation of radiologists/new grads (with 300k$ loans) that will be caught off guard because you know...it's hard to adapt when you've committed 8-12 years of your life learning one skill. Are you going to help them transition to becoming seamstresses? Even changing specialty would be a stretch. Hence, why we are here trying to make the best educated guess; nobody wants to be part of that group that gets sacrificed aka the coal miners aka "buying at the top." It's a very valid concern.

Again, no one is arguing about the greater production and increased efficiency; that's all good and dandy.

 

[Dro133]

Maybe I'm way off the mark here. My thinking is that while we're still far away from "Watson, interpret this CT scan", AI is rapidly getting better than humans at answering questions like "Is there X present?" Is there a pneumothorax, is there a head bleed, is there a lung nodule? It's hard to predict what the downstream implications of this will be as AI continues to improve.

My understanding is that with the current state of medicine, basically the reason we don't pan-scan people (cost aside) is that we find too many incidentalomas that need to be investigated.

I think it's possible that due to the ability of AI to consider so many factors that the human eye cannot, we could see major paradigm shifts in screening and the ability to tell benign vs malignant based on imaging alone. There's an entire new area of study called radiomics, which has only existed for <7 years, dedicated to interpreting the nearly limitless number of "imaging biomarkers" and statistical relationships that can be extracted from imaging.

People apparently thought CT was going to make radiologists obsolete because "it's so easy, everyone will just read their own scans" - look how that turned out. I think that as AI gets better and better, the amount of information available is just going to exponentially increase. Clinicians will want and need help with interpreting that data and figuring out how best to apply it for each individual patient.

Or maybe not. Who knows

I agree with the main premise of what you're stating. I think AI can eventually be at least as good at radiologists at answering "is X present". I also agree that, more importantly, AI will allow us to use imaging features (or biomarkers) that are not visible to the human eye in order to improve prognostication and treatment response prediction. My interest in that space is essentially why I'm considering a career in pathology myself.

I guess I can see your point regarding why these advances could lead to pan-scanning becoming a viable practice -- i.e. if we can use imaging biomarkers to differentiate benign vs. malignant with very high confidence, rather than needing to biopsy, maybe there aren't as many downsides to pan-scanning. But this would require 1) very robust radiologic-pathologic correlation and 2) really good validation studies to confirm that there is indeed a net mortality/cost benefit to this practice. Interesting stuff though.

 

[YoungDoc25]

AI alone will greatly impact or replace DR but the fields rampant with midlevels I’m sure will be completely safe when AI reaches that level of understanding

 

[cecumbowels]

I dont mind the intrusion. However, like I mentioned earlier I am not worried about AI and its affect on the future of radiology.

 

[nimbus]

One of the basic questions is at what point will we trust a system without a human at the wheel?

 

[radsisrad]

One of the basic questions is at what point will we trust a system without a human at the wheel?

Well... how many humans need to be behind the wheel?

Radiologists are certainly experts in imaging, but we are not the only physicians on the ground - there are EMs, internists, FPs, subspecialists, etc. What if (when?) AI gets so good that residency training can fairly seamlessly incorporate imaging training in the given specialty - e.g., neuro residency includes AI-assisted education in neuroimaging, Heme/Onc fellowship does the same for oncology imaging, etc.? Does an EM doc need a radiologist's final say on suspected aortic dissection or PNA, or is AI's input (along with the MD's clinical impression) going to be enough on which to reasonably act?

For the foreseeable future, we will need radiologists... but the odds are that we won't need as many once medical AI gets rolling in earnest. If not now, radiology leadership needs to be poised to slash spots on a dime at the first signs of significant inroads.