Dear Dr. Cellini, Naglis, Chief Medical Officer and researcher from Oxipit (also a radiologist) here, found your excellent channel, an honor to be featured! First of all, thank you for an excellent video and your valid and reasonable points. Let me provide some feedback: 1) Software made no (clinically significant) errors during the pilot phase. Your comment was correct that it does not list what was considered a non clinically significant error. Let me elaborate briefly about this: clinically significant errors are defined as errors which if unreported have the potential to cause patient harm. This would include everything that would be broadly considered clinically significant and/or actionable (nodules, pneumothoraces, consolidations, subtle minor nodular opacities, which cannot be excluded to be significant and many others, which are too many to be listed). Some findings such as degenerative changes of the thoracic spine, some age correlated sclerotic changes of the aortic arch and other similar ones, can be considered clinically insignificant and might be classified among the normal studies. There will be findings, which might depend on the institution if they are to be considered as significant or not, and this is identified during the initial retrospective and prospective validation stage in each institution. Any steps towards autonomy are only made once the radiologists from a specific institution are confident in the validation and safety of the solution. 2) Who is liable for the AI software? To answer this question shortly, once the software is running in an autonomous mode we aim to take the responsibility for the analysis of the studies, which were classified as normal by the software. You have correctly noted, that if that is not the case, there is marginal value to be gained for the radiologist, as at most this will save only a few seconds of time per study. Before we get to the autonomous stage and the final signature is still on the institution/radiologists side, the liability is the same as it is now - on the institution side. Having said that, there are questions which we cannot yet fully answer, we aim to be transparent about that. Even though we are already certified, there are national and international laws which are still to be worked out. CE certificate is not the end of this process - it is a benchmark which allows the next steps to be taken. 3) Will this actually help radiologists? How much it will help exactly depends highly on the institution and the market. A few examples: A primary care institution which makes 100 CXR studies per day, and lets say on average 80 of those studies are normal. Ultimately the software could potentially automate the reporting of ~35 of the studies, which could result in tangible resource/cost savings and allow the radiologists to spend more time on more interesting and challenging cases (either CXRs or other modalities). Also, there are institutions in some markets who already cannot report a significant fraction of their studies (both in western and the developing world countries), and some patients do not get a report for their chest X-ray either altogether, or during a useful amount of time. We believe automating even a sample of these studies can significantly contribute to improving patient outcomes. 4) How much does the software actually cost? Even though I cannot give a specific price per study (this depends on the market and the cost per CXR in a specific market) this is projected to cost less than the radiologist resources which would be required to report the automated studies in a regular way, with a significant margin. Finally, in regards to your finishing notes about if the radiologists are at risk to lose a significant amount of work anytime soon: short answer - no. This is the first step to automate a fraction of the normal studies of one radiological modality, volumes of which are already too large for the current radiologist resources to properly manage. It is likely that somewhat soon we will see some similar examples in a couple of other modalities. However, I do not think that it is or will be possible to automate abnormal or borderlilne studies anytime soon, mostly because of factors not related to the AI performance, but because of radiological factors, such as the importance of clinical information/context and interreader subjectivity. For the obviously normal studies the effect of these factors can be minimized, but for abnormals they will remain very substantial. Therefore the large majority of the radiological studies will remain in the radiologist's hands for the foreseeable future, unless we figure out general AI soon, which will take all of our jobs and create an utopia on earth. Thank you once again for the feature and if you would like to discuss further, we are open to your questions :)

And then in the fine print: "We (the super-mega-smart AI start-up) don't assume any risk or responsibility should misdiagnosis by our AI software lead to a serious health consequences. Every final clinical decision rests on physician's shoulders."

Hello Dr. Cellini, As somebody pursuing a PhD in medical image analysis (the branch of AI that applies AI to problems in radiology), I agree with your points. Artificial intelligence is unlikely to replace radiologists any time soon and we could probably use more radiologists rather than fewer. In my opinion, AI will not render radiologists obsolete for several reasons. First, the performance of the algorithms depends very much on the modality that is being used and the image analysis task that is being considered. For example, if you want to delineate thrombi on NCCT/CTA with an algorithm, this is a very difficult task due to the object being very small, image artefacts making the scan less clear and sometimes no hyperdense artery sign being present. If however, you want to delineate very large tumors in the brain on MR scans this is a very easy task because they are large and easily visible. The latter problem would be a fairly easy to publish research paper and would draw attention from the media, but not the former problem. So this creates the impression that AI is good at everything, even though it is not. Second, there is a problem in AI that is called the domain shift problem. To put this into a medical context, if you train on a dataset that was acquired using machine A and you report your results on data acquired on the same machine it will get a certain performance. If however, you change the machine that you use, it will get a lower performance due to the images looking slightly different. This would render some algorithms that work well on relevant problems useless until they are re-trained on data that includes scans made on this new machine. Third, the data leakage problem. Many early research applying AI (deep learning specifically) included people that did not have much experience working with medical data. These were people coming from computer vision where you work with individual images (most of the time) rather than patient data. As a result, specifically CT and MR scans were not correctly distributed to training, validation and testing sets. The correct way of doing so would be assigning patients to splits. What would happen was all slices of all of the the scans would be shuffled and assigned to splits. This means that highly correlated slices would be trained and tested on, leading to inflated results in some studies. Now to provide some context, I got this information from a colleague working in the same lab as I am and did not read the original paper, but I have seen this problem in a number of studies myself. Fortunately, this is improving with better practices and the field maturing more. To end on a positive note, there are several areas that, in my opinion, AI will be very helpful for. A great example is research. If an image feature is labor intensive to annotate but could be a possible alternative endpoint AI can definitely be of use here, especially when you are working with large-scale clinical trials or registries. This makes research into the efficacy of medication and treatment much more feasible. Another application would be to allow for quantification of image features that inform treatment or are prognostic, but that are too labor intensive for a radiologist to do. Volume measurements are a good example. A third line of research that I find quite promising is that of image reconstruction. Reconstructing an MRI image from K-space is very time consuming. Thus, in addition to MRI having several contra-indications, acquisition and reconstruction time are limiting factors. There is research that is working on requiring less samples from K-space, which speeds up acquisition. Finally, AI can help make the image data more easily readable for a radiologist. For example by registering images more closely or extracting center-lines to "stretch" out arteries. Hope you find my response helpful. Kind regards, Riaan

I just wanted to say how much respect I have for IR doctors. I do have Crohn’s disease and I became very sick last week. I’ve had a non healing abdominal fistula for a year now. I’ve had 3 drains put in the past year. Last week one of my old drain spots started draining and along with one of my fistula holes on my stomach. So I went to hospital and they found a huge pocket of infected fluid along with stool from the fistula being connected to my small bowel. I also have an ileostomy. IR did place another drain in me but I’m going to need surgery to fix this fistula. But the IR team were amazing. I really appreciate what you guys do.

EKG is a great counter point. I’m an Emergency Med resident and the machines are notoriously wrong, both with over reading and under reading. Unless it’s stone cold normal the machine read is useless.

Excellent. As a radiologist for the last 40 years- I have been told radiologists will be replaced by other specialities and AI. Still practicing with no real fear. Your assessment of the situation is perfect. We need to be involved in the development of the AI technology for DI and in incorporating it into the radiology diagnosis algorithm. Great explanation given by Dr. Cellini.

Hi Dr. Cellini. Love your videos. They made me interested in switching from xray to IR tech👍. Keep up the awesome content

Thanks for the grate channel that adds a lot to us. I’m a PG-Y 3 radiology resident, who just got a chance of having a scholarship for emergency radiology fellowship after the residency. However, right now the only subspeciality i think about is intervention. Should i accept the scholarship for emergency radiology or take a chance to find someone to offer later on “may be” , your opinion is needed. Many thanks :)

Also, you are 100% right about the EKG software. Our department chair in radiology told us that studies at our hospital showed that the software is wrong like 50% of the time. In other words, never follow what the program says. Read it yourself.

Hi Michael! It's the same with CPAs! I think AI and advanced technology may have an impact for accounting technicians in the future, loosing their job?? We will see!

I really wish AI companies would start focusing on making me more efficient as a starting point. For example, have the AI program pull out clinically relevant information from the chart or automatically measure a lesion that I select (which would also improve consistency). These steps would be a great spot to start and could also allow for data collection which would help further AI development in the future. We use basic AI to diagnose bleeds, PEs, etc. It's definitely good for prioritizing workflow but it makes a lot of mistakes. Would be terrible to rely on it.

I'm not a radiologist, but a machine learning practitioner, so I'll give my perspective. The issue with radiology is that that problem is in principle an ideal "shape" to be tackled by machine learning. The job involves a large amount of transforming data from one form (imaging, patient data) into another form (language), and there's plenty of potential training data being collected every day. In machine learning, progress is often sudden. A problem will go from "almost impossible" to "mostly solved" in a short period of time without much warning. For example, the first computer to win a professional Go match against a human: October 2015. The last human to win a professional Go match against a computer: March 2016. 6 months. A couple of years before, the conventional wisdom was that that this problem was at least a decade away from being solved. If you are planning a career, not just in medicine but in any discipline, and you want to know if your career is at risk of major changes as a result of machine learning, it would be a mistake to look at how researchers are trying to solve that problem today. Few of the methods we use today will be in use 10 years from now, let alone 20. The primary architecture of modern language models was only introduced 5 years ago, for instance. Much of the language modelling research that was done >5 years ago is obsolete. The fact that a problem has not been solved yet means that the state-of-the-art is wrong. Not that the problem itself is insoluble. Instead, you have to look at the fundamentals. How easy is it to train a model? How much interaction with the physical world is essential to the job? If much of the job involves doing things at a computer, the risk is increased. Will human radiologists still exist in the future? Yes of course. Will ML models *one day* gain superiority over human radiologists at a majority of relevant tasks? Certainly. Will this shift occur within the careers of today's radiologists? It seems likely, given the rate of progress in related areas.

love to see AI take the core exam before being allowed to practice

Awesome vid. I'm a radiologist from the Philippines and I study ai for radiology. Just had to ask. What monitors do you have in the background? They look great!

As a dentist studying computer science, I am fascinated by this topic; genuinely hoping to see AI applied to orthodontic treatment planning. There is so much good data now with many orthodontist taking pre and post tx CBCT.

Question for you. I've always been taught, even as a third year, to always read through every image I order as a future resident and physician. And more importantly, it's a bad idea just to read imaging without having any idea of the history or physical exam findings to compliment the imaging. It doesn't seem like this machine does that. It seems to me that will be a major limitation of all AI programs with regards to imaging. There are just things an AI won't look for if you don't program the relevant history findings into it. What are your thoughts?

Could you make a video going through some radiology cases you found interesting or scans that are complex so we can see your thought process on these scans

In the past, when computed tomography was the big new step, they sayed that the radiologist would no longer be necessary, after all, they thought it was possible to see the inside of the patient. What happened was the opposite. The radiologist now has a great new method, greatly increasing his work and relevance.

Radiographer here!!! I have a legit question…do you guys actually read our exam notes in the patients chart when we put them in? Also if you actually do you you like more details or less? Like mechanism of injury, or a more thorough? Like hx of chf, asthma’s, etc? Always wants to ask my rads but too scared of seeming ignorant! Thanks!!

I would be interested to see Dr.Tommy thoughts. With you both being at different stages of your careers it would be interesting to see both of your thoughts and views on AI.

I would make 2 arguments for AI. 1. The double check is awesome. While you suggested it's a doubling of work, it's also a bit of a failsafe. Redundancy in this field is beneficial. 2. AI could be used to prioritize reading. It's like a computer based triage. Get the radiologist looking at likely abnormal images first. Delaying care for 80 no-finding reads seems unnecessarily risky. I work in a Cardiac Cath Lab but my most significant experience was as a patient post-covid... An emergency room Doctor started treating me for blood clots and DVT's before any radiologist saw anything from my medical record because it was so obvious when the images were in... I can imagine in some places that wouldn't have happened.

Hello! Do you think the use of AI could lower the salary of a diagnostic radiologist, since it will lower the amount of work they do? I am interested in diagnostic radiology and wanted to know if this will be a problem.

How is this different from any diagnostic test in terms of liability? diagnostic tests have published sensitivity and specificity, and clinicians are not liable when the test is incorrect. Also, this is just one of many such tools being developed. Would be interesting to get your take on how ai will impact your job in 10 or 15 years and what types of tools would be most useful.

I once heard somewhere that AI will never be autonomous because it'd be very difficult to sue when mistakes happen... and we love to sue.

Thank you so much for this video! Radiology is my dream but I have been worried about AI and this has cleared a lot of my fears!

A lot of very good points mentioned! In my opinion the strongest argument agains any consistant use of AI in medicine is accountability. Since the company is never going to say "our software is 100% correct" and definitely not even consider making themselves accountable for the mistakes, their argument will always be "human is also not 100% correct". Which is true. But physicians and hospitals would (not to say never) but most probably never go into that direction and even try to risk it - there would always be a second glance at the study, and this is where it loses its point. By far the best comparison is ECG: anyone who spent even a day in the ER never took a look even at the value of heart frequency...and as you said "these are *lines* ".

As someone applying to Med school next year and looking to Radiology, I agree that I am more excited about AI than worried. Additionally, the whole “AI taking over” really only would apply to a pure Diagnostic Radiologist, as like you mentioned as an Interventional Rad you do much more than reading scans. Great vid on the topic :)

Hey Dr. Cellini, how do you feel about radiologist assistants? They can also help free up time for radiologists by taking care of minor invasive procedures and fluoro cases.

I read EKGs which are infinitely easier to analyze than a CT scan or other imaging. And even with those, the computer analysis misses critical diagnoses regularly, and get minor things wrong 20-30% of the time.

CE mark is actually more than FDA approval. It's a general certification that it meets all the requirements for being put on the EU market. It could be health related legal requirements, or for electronics and other things it could be more similar to UL/CSA safety rules plus rules about chemicals used in the product. Lots of other requirements for other product categories

AI will "replace" Radiologist the way calculators "replaced" mathematicians. Technology amplifies human output. Ironically, when the x-ray itself was invented, "people" thought it would replace Physicians entirely. Why do we need a Physician when we can just look at these images and know exactly what the problem is? It turns out x-rays and other forms of medical imaging vastly expanded our understanding of how little we know, and the need for well trained Physicians actually expanded due to the need to interpret these ever complicating technologies. AI will likely be implemented in Physician training as an augmenting instrument in delivering patient care.

Hey Dr. Cellini I am a software engineer/computer scientist and have worked with AI in the medical field and implemented a lot of software. Your points are correct: AI is not good enough to replace actual doctors, but it can help them in a lot of things. Maybe in the future we can rely more on it but definitely not right now. I really doubt that it can replace a doctor 100% but we will see what the future brings us

Im a data scientist (the people developing AI software) and play around with the idea to go back to university and study medicine. And the most interesting area is actually radiology to me :)

I find it interesting (and maybe somewhat telling) that in the list of benefits that AI may bring you didn't even consider the benefit of AI checking your work, rather than you having to check its work :-) I think that is the ultimate pacient perspective. If someone told me that the doctor's work had been check by a tool that has been designed to avoid some of the human biases (such as anchoring), I'd be quite relieved. It's been some time so I don't remember the exact numbers. But there was an article in Scientific American on a study where they compared performance of a solo nephrologist with a nephrologist aided by a computer. The doctor + computer tandem had a success rate of 60-70 per cent in diagnosis, where the standalone doctor got the diagnosis right only about 8 per cent. That should be absolutely shocking and game changing finding, yet, it was only a side note in a popular science journal and the world keeps turning ...

hello dr , im new here but I have a question for you maybe. You can answer that for me , im looking into starting med school and radiologists is been catching my eye in becoming one. Do you think? it will still be worth majoring in radiologist do too AI?

Future medical physicist here! As you stated,AI won’t be replacing radiologists anytime soon…I’d even argue at all lol. I do research in adaptive radiotherapy and while the software is very advanced + increases the precision and personalization of cancer treatment, human analysis is still superior and necessary to correct for mistakes in the algorithm😊. I always find something to correct in the contours and auto-segmentations made by the computer. With that being said, AI is still super helpful in speeding up the planning + treatment process.

Hey Dr. Cellini I can't believe you really did a rotation in Malawi. I'm Malawian and sadly we still only have like 3 CT machines the whole country sob sob

As a pathologist, we are beginning to see the incorporation of AI into daily practice…it’s very limited right now but I’m confident it will explode over the next 5 years. It will revolutionize pathology and will probably make us more efficient. I wouldn’t trivialize how normal chest X-rays can be read by AI algorithms because it’s just the start. My philosophy is we have to work with them instead of against them….our jobs aren’t necessarily at risk but they will evolve. Just my two cents!

Good point Doc! One thing, however, I worry about AI deployment for use in developing nation is that AI is basically a machine highly optimized to generate nearly accurate impressions for report based on inferential logic analyzed from vast data of prior diagnostic images or scans it was fed from the Western or European population's pathologies which in some cases are inconsistent with developing nation's pathologies. I am not talking about the obvious consistent pneumonia, cancer etc..., but things like internal artifacts from foreign bodies since most developing nation populations are dynamic and often in direct contact with the environment. Growing up in the village, I always remember walking bare foot and piercing from splinter, stepping on ails, falling on sharp woods from a tree, solid food stack in the throat, solid fruit seeds in the stomach, lung damage from windowless room smokes and so many more that are not obvious in the Western/European world. Thus, isn't it possible that the AI's accuracy in rendering relevant impression becomes questionable since it does not know anything other than making educated artificial guesses based on the Western/European diagnostic data it was fed?

Tq a lot for this video i was really in a dilemma about taking up radiology n this helped me

I would think this is also assuming the proper technique was used and the image is not over or under exposed. I wonder if it would cause it to need to be read by radiologist if it wasn't performed properly

For at least 12 years GE software for nuclear stress test has been able to give an interpretation of the gated SPECT images. The scary part is I have seen a radiologist use the softwares interpretation verbatim in their dictation.

2:15 I'm very curious how the programers would react to something like that happening in their field, what will they think, say and do if there will be some initiative to fully automate their jobs . Don't get me wrong, thigs like this will make (almoust) everybody have a better quality life, but I hear the "fully automation" of every job , except those who make those algorithms ...

Important to remember that AI is still in its nascent stages. ChatGPT is akin to a single cell organism. They will grow and grow very fast in complexity over the coming years especially if no restrictions are placed on research.

Nice video. I am an MD and also a software engineer. I lead many AI efforts on FDA cleared products and research projects at universities. I 100% agree with you. AI is not replacing radiologists anytime soon. The AI state of the art is not there yet and I doubt we will get there without a fundamental change in AI tech. This is why "self driving cars" has not reached Level 5 autonomy yet. The technology is more like glorified pattern matching than being able to "think." I think the most valuable application of AI in radiology is helping the radiologist perform mundane tasks instead of trying to replace human cognition. There also are other applications of AI like AI driven CT image reconstruction and AI enhancement of MRIs-allowing faster scans with similar image quality. I do think the EKG example is valid. However, I would question how new those EKG readers are? I think recent advancements in AI and ML-particularly deep neural networks-have improved those algorithms significantly.

Would you be able to do a video on teleradiology?

Hlo doctor, will you suggest radiology for a medical student who has dry eyes

Perhaps AI could raise red flags to prevent misdiagnosis - had it caught my erroneous reading it may have prevented going thru medical hell, almost killing me and shortening my lifespan

Tbh our transparency is largely underestimated and potrayed in a flawed manner. It's a pity these companies keep everything under a * . Despise how they hype up, get hopes up & then everything goes plummeting down. They should make an attempt to increase accuracy before hyping things. Pardon me for the bitterness. I'm in fact FOR AI. I just feel healthcare software needs to learn how to take responsibility. I loved your analogies, Dr. Celini. Thanks for the insights

AI should not be used in the direction of getting reports, instead it should babe used for getting clearer and enhanced high resolution imaging. I can easily see how AP, Lat L&R and PA view XRays can be combined to give a some what 3D images, that would be awesome.

I have done research on this before. The black and white images are not easy to analyze using a ML model. I think the ML tool will help catching problems that are not noticed by radiologists.

I am much more curious in how AI increasing a radiologist’s efficiency will impact the supply and demand of radiologists. For ex. If normally it takes 10 radiologists to clear a typical days imaging workload. Will Radiologists+ AI increase efficiency so that the same work can be done with 6-7 radiologists? If so, why wouldn’t businesses employ AI to increase efficiency and cut overhead costs? I think this is a much more reasonable role for AI in the short term because it doesn’t rely on the technology being perfect. Even if it can do simple things such as measurements or early alerts on potential findings, it could still drastically increase efficiency wouldn’t it? Overall it seems this is better for patient care, decreasing inefficiency in the system. I guess the real question is whether the increase in radiologists efficiency with AI will outpace increased imaging demands. If that does happen, maybe groups can leverage the increased efficiency to expand coverage area, maybe by employing teleradiology? What’re are your thoughts?

This video has a ton of good points. Strangely enough, I know a guy from college who tried to start a company doing AI radiology as well lol. Just like you said liability was a huge issue, and he couldn’t get a hospital to even consider his program for that reason. They were not willing to risk a lawsuit over the AI reading being wrong if they solely relied on it. And even if they did do that a radiologist would have to read over the chart themselves to confirm a diagnosis which defeats the purpose of having an AI reading software in the first place. So thus his company failed no surprise. I think Radiologist careers are safe for now, until we get fda approved AI robot radiologists way in the future 😂

Is this talking about replacing radiologist who diagnose or the techs that take the photos or both?

For me I would have a board certified radiologist read the cxr. Then put them through the AI to see if the radiologist missed something. We are all people and we can make mistakes. Having a second look doesn't hurt but makes mistakes lessened

I feel like it would be good for doing tuberculosis screenings in areas with high TB infection rates?

I think it could be a subspecialty to the rad tech license in the future, where they could double check the x-rays that the AI said were normal. That could be optimistic but it's an idea?

Don't worry boys and girls you can safely go into Radiology without having to worry about robots. We become one with them. As you said, they aid us not completely take our jobs.

I suspect there will be a role for AI - so far technological improvements have been great for patients but have made the job of radiologists infinitely more difficult (CT scans going from 20 images to 2000 images for example). But even the technology that improves workflow (going from typewriters to word processors for example) don’t devalue the role of the person using the tech.

Hey dr cellini would you please make video on alternate radiology pathaway for IMGs.........

I am now tasked by the Karolinska Insitute and Stockhom Universsity to research about diagnosing based on CT, MC and Xrays by AI and statistical methods.

Would love a video of turf wars in diagnostic and interventional radiology vs other specialities.

Does it say something about the current state of medicine the first thing we discuss is liability?

Good topic and observation- welcome AI where it can reliably benefit radiology (it ain’t goin away) and know your click bait. I can only relate such a topic to US and can’t quite imagine AI interpreting beyond WNL. Except, in time.

What is “The contrast appears to be have progressed to the distal small bowel loops.” Mean on CT scan results?

is is legal for a ordering practitioner to put in a order prior to seeing a patient?

AI Company: What do you want? Radiologist: Software that removes redundancy in my workflow AI Company: OVERREADS IT IS!

At least it helps to prioritize the CXR, so Radiologist reads and reports the problematic x-rays first and then validate the normal looking AI results at the very end.

By the time AI fully replaces a radiologist, there will be plenty of other medical specialties that will also be replaced. There’s a ton of interaction with the clinical services and integration of clinical context that goes into the interpretation of studies that require a level of AI that just does not exist. Unlike what most people think, image interpretation does not happen in a vacuum.

What I dont understand is, why radiology specifically? Like why aren't people worried about AI replacing pharmacists, PCPs, Heart Failure Cardiology, Diabetes Endocrinology, PAs/NPs or even ED? So much of medicine is algorithmic, "bread and butter" stuff. If you have the technology to read imaging data then you have the technology to read basic labs, ekgs and salient features in a H&P. So much of radiology is correlating the image to the clinical picture of the patient it just doesnt seem possible that AI could *replace* the physician. The only reason people have been saying this about radiology for a decade now is because radiology is driven by and interested in innovation. This is only going to help.

The clinical model for this already exists in mammography. Everything has to be double checked by the radiologist and AI more often impedes rather than aids. And it has been this way for, literally, decades. AI can't be just as good as the radiologist, it has to be a standard deviation better and across different institutions, with different artifacts and fields of view. This is why that guy who won the Turing prize resigned (read: was fired) from Google. I suspect this is just another drop in the Gartner hype cycle bucket.

For MRI AI software is already here for GE. The AI software basically makes the scans faster, you could scan a Lumbar in about 5 min which means instead of scanning 26 today maybe we could do 40 hahah I could see helping the Radiologist and reducing workload but replacing them yet. I've definitely seen Radiologist miss pathology. It's understandable, especially if the rad has to read over 80 scans a day.

will Radiologic Technologists / Radiographers can also be automated?

There's a guy in a TED talk I've seen recently saying we should stop training radiologists at all, right now. He thinks that in the next maybe 10-20 years max. no radiologist will be needed. If the algorithm outperforms radiologists significantly (and there are some AI's that do that already for certain tasks), there is no need for them anymore - and the hospital will happily take responsibilty (they do for radiologists as well). I do not share that opinion (I think it will take a much longer time), but I think the demand for radiolgists will decrease and their salaries will suffer a lot. Actually AI is better at Chest X-Rays than it is at ECGs because an ECG is funnily enough more complex than a chest-XR (which is in the end just pixel-intensity on a grayscale in a certain location while an ECG-line carries multiple information like time, voltage and therefore location - 3D-vector projected into 2D - and that from different positions from all the different leads)

Yeah, but wouldn’t a actual radiologist have to input data for the program/computer to work?

As a fresh MD and having a background in electrical and computer engineering, I find the future of radiology (and medicine in general) very exciting. I am for sure going into radiology 😁

As someone who works in AI and Modeling for cancer diagnosis. I can say that we’ve fed millions of images through our AI for cancer diagnosis and the AI has learn to detect cancer very early, however there will always be the need for human review of the images.

What was your pre med major

I fly like the CXR reading AI is kinda like what we do as AEMTs for my job. No one trust the ECG auto interpret so our job is basic ECG interpretation. We are trained just enough to identify rate and rhythm and if anything doesn't meet picture perfect we have a paramedic look at it.

We need AI for hospital administration

Surely the company has a legal disclaimer to the effect that all AI readings must be confirmed by a human rad as to ensure the malpractice suits against MD's don't stop

And also..are the rads familiar with what it entails to get images? Sometimes I have to send through sun optimal images…(which I absolutely hate) but it’s not always easy getting images on pts that are fighting you or have dementia. I had an Ed doc come over a couple years ago because a lady had dislocate both elbows and he came to see what was taking so long. When he saw what we were dealing with he was like “ohhh shit. I had no clue!” She was screaming and trying to kick us. No fun!!

Do you follow medschool insiders ?

U can look at chest X ray and if it's ok that's 10 seconds but is not then is like 1 minute as u must write a report and that's 1 minute if you have a good report template so this software is not bad at all, could replace a radiologist, i don't think so.

The real Q for medicine and patients is, how much is a missed diagnosis worth? If the settlement cost is less than that of hiring radiologists and other factors on the opposing side of the equation, you have your answer. At least, I imagine, in the US healthcare system (jks).

I dont feel like we will be replaced, however I disagree where You say there is no point in the software. I do feel like that every study even cxrs should undergo double reading. Now realistically there will never be enough radiologists for that ever, especially with the way volume is increasing. I do feel like this is the second best thing we can have and we should utilize it. Sure it will not necessarily help with volume, but it will surely improve with quality. At least thats one area where I can most certainly see AI be a useful tool, tool beeing the most important word. Of course if You just be expected to sign unread reports that would be mental, and no radiologist would ever agree to that.

as a radiology resident, I ca'nt deny Ive got kindy scared after the main tittle. Greetings from México

Pathology too... I read a few Journals on it.

Great video, thank you

Given a set of 100 images machine learning was outperforming radiologists 15 years ago by 7 percent margin or so. Today with more powerful neural nets and AI - (which have been around for 25 years and doing voice recognition since dragon speak) the gap would no doubt be wider . Given a large training dataset there is no way that a human radiologist would outperform the machine - the reasons they have yet to replace radiologists will be to do with other factors than ability to diagnose and not necessarily in the patients interest

The important point is AI has got its foot in the door. If we look at the history of AI such as chess, GO, and Poker it is not a question of if AI can become better than humans but when. In medicine, it is finding the right AI algorithmic system such as by doing "search" to improve capabilities.

As per GDPR, AI should not make automated decisions without human intervention. So the AI is to assist radiologist and provide a recommendation but the final call is of a radiologist.

Well, everytime someone tells me that the AI will still my job I just laugh! There's so much more complexity in reading exams that anyone can't even imagine!

AI will replace not just most radiologists, but most jobs that require using your brain to identify and calculate. What it will have trouble replacing is jobs requiring creativity. Maybe not the next 5 or 10 years, but eventually, algorithms will achieve superior understanding of imaging than humans, so much so that it will be unsafe for humans to read studies. Look at self driving cars for example; we're nearing the inflection point where self driving cars are much safer than human-driven ones. I do agree with the point that AI will encourage more creative thinking in radiology and other medical fields. If you have superhuman diagnostic skills available to you, then you can maximize your problem solving skills with unprecedented tools. The future is exciting!

I Looove your videos. I'm tired of hearing my teachers saying AI is going to take radiologists places. Usually people who say that don't know anything about the subject hahahahaha 🇧🇷🇧🇷

I can see how so called AI can be of assistance to a radiologist in certain ways like outlining or coloring specific areas or following a complex structure or vessels/nerves across images but in general this frenetic enthusiasm for AI in medicine that goes back to at least the 70's has not panned out in any meaningful way in clinical practice for some of the reasons as stated by this obviously very smart and articulate professional radiologist. The software and other engineers insist on foisting all manner of stuff on the medical profession that often is not practical or actually gets in the way. Number one on that list is the electronic medical record systems. All kinds of grand ideas and enthusiastic claims but in day to day practice just stupid junk that is a major pain and slows medical care making it far less efficient that just jotting a quick note or fast dictation. There is so much additional imaging being obtained - mostly not needed, just adding confusion and steering patients to unnecessary potentially dangerous care that radiologist will not be threatened by so called AI, they will continue to thrive as they shovel loads of BS looking t=for the occasional medically legitimate difference maker. As for Africa , the situations there are often ridiculous not because the people are not fine people (which they are ) but because there is often severe and systemic corruption. There is no country that should not be able to well afford whatever # of scanners that would make sense. Its just that the money to do so is going elsewhere. Likewise if many of those countries would spend less on training soldiers in lieu of radiologists and other MDs their health systems could begin to rival what should be expected. If they need defense they should instead enter into far less expensive defense alliances with the US, UK, the Dutch, etc. (but not the Italians ;-) As for me i love the radiologists, they are a big help and if i want to be humbled i go to one of their conferences and see how little i know.

My best friend who is an engineer came to my house and asked what a pneumothorax is, and what it looks on an XRay. Few days later he came with a laptop showing me a software that is better than most doctors in detection of Pneumothorax.

I cant wait for the automated Hospitals like in the movie Idiocracy. LOL

As a radiology technologist this is super interesting. And to say that radiologist can’t read a normal chest X-ray in 10 seconds is false as I can look at an image and know if it is normal or not within 2 second it becomes second nature

It’s not a sad day, many radiologist miss a lot of stuff- also there is a huge shortage