Will X-Rays Soon All Be Read By AI?, with Louis Plesner

Episode Overview

Episode Topic: In this podcast episode of The Skeleton Crew, host Jen Callahan interviews Dr. Louis Plesner. He is a remarkable medical doctor guest from Copenhagen, Denmark. Dr. Plesner is also a Ph.D. student working on the application of machine learning in radiology, which could have an impact on the X-ray Tech role. He is also contributing to an emergency room and teaching ultrasound techniques.

Lessons You’ll Learn: Dr. Plesner’s journey from a radiology residency to specializing in machine learning for radiology applications sheds light on the evolving landscape of medical imaging. He emphasizes that radiologists, including X-Ray Techs, should not fear technological advancements like AI, as they complement rather than replace the role of human physicians, especially X-Ray Techs. He also highlights the potential of point-of-care ultrasound in various clinical specialties for its versatility and patient interaction benefits, which is particularly important for X-Ray Techs.

About Our Guest: We are excited to introduce Dr. Louis Plesner, an accomplished Ph.D. candidate in the realm of radiology. Dr. Plesner delves into the dynamic intersection of machine learning and medical imaging, shedding light on how advanced technology can revolutionize the interpretation of radiographic images, benefiting X-Ray Techs as well. His passion for patient interaction has guided him to a pivotal role within an emergency department, where he lends his expertise in ultrasound techniques to enhance patient care, including the care provided by X-Ray Techs.

Topics Covered: Within this episode, Dr. Plesner unveils his distinctive voyage transitioning from cardiology to radiology, narrating his immersion in the realm of machine learning. The discourse navigates the convergence of AI and machine learning within radiology, underscoring their role in amplifying, not supplanting, the duties of radiologists and X-Ray Techs. Dr. Plesner elaborates on the efficacy of natural language processing models such as ChatGPT, unraveling their scope that extends beyond radiology and benefiting X-Ray Techs as well. Envisioning the times ahead, he contemplates the potential of AI in automating reporting procedures and elevating patient care standards, including those upheld by X-Ray Techs.

Our Guest: Dr. Louis Plesner, Medical Doctor and PhD-student at Herlev/Gentofte Hospital

As a driving force in the ongoing evolution of radiology, Dr. Plesner’s expertise in machine learning and radiology places him at the forefront of innovation. His research explores the potential of AI to enhance diagnostic accuracy, streamline workflows, and improve patient care. Dr. Plesner’s commitment to both academia and patient-centered care exemplifies his dedication to advancing the field of radiology.

We are thrilled to have Dr. Louis Plesner as our esteemed guest, eager to delve into his insights, experiences, and the transformative impact of his work in AI and radiology, impacting the daily life of X-ray Techs.

Episode Transcript

[00:00:00] Louis Plesner 

I want to say to all junior radiologists like myself that don’t be afraid to go into this extremely cool and nice speciality just because some people are saying that we will be out of a job, I don’t think we will, you can never predict the future. But if radiologists will be out of a job, I would say that all other physicians will be out of a job as well. The algorithms would have to be that advanced to put all the physicians out of jobs and that could happen. But you cannot really think that way because then we would probably just do something else. 

[00:00:33] Jen Callahan 

Welcome to the Skeleton Crew. I’m your host, Jen Callahan, a technologist with ten-plus years of experience. In each episode, we will explore the fast-paced, ever-changing and sometimes completely crazy field of radiology. We will speak to technologists from all different modalities about their careers and education. The educators and leaders who are shaping the field today and the business executives whose innovations are paving the future of radiology. This episode is brought to you by xraytech.org. If you’re considering a career in X-ray, visit xraytech.org To explore schools and to get honest information on career paths, salaries and degree options. 

[00:01:21] Jen Callahan 

Hey everybody. Welcome to another episode of The Skeleton Crew. I’m your host, Jen Callahan. And today I have a remarkable guest with me today. His name is Louis Plesner, and he’s joining me from Copenhagen, Denmark. Louis, do you want to say hi? 

[00:01:35] Louis Plesner 

Hi everyone, and thanks for inviting me to the podcast. 

[00:01:38] Jen Callahan 

Oh, my pleasure. So just to give you guys a little bit of a background of Louis, he’s currently a Ph.D. student who is working on machine learning for applications that can be applied to radiology. And he is also currently working in an emergency room doing a little bit of teaching and then also some oversight of the department. So I’m going to let Louis really give more information on that. And then we’re going to delve into his whole world of radiology and then into machine learning. So, Louis, just give us a little bit of background of yourself. 

[00:02:14] Louis Plesner 

Yeah. Thanks a lot. Again, thanks for inviting me to the podcast. Really delighted to be here. So, my name is Louis Plesner and I’m currently a Ph.D. Student in radiology working in a quite large hospital in Copenhagen in Denmark. I’ve been working on this Ph.D. Project for two years and some by now, so I’m starting to get into it. And I joined this project after my first year of radiology residency, which I also did at the same hospital where I’m currently doing my research. I was in the residency when Covid hit us in Denmark and I was doing thoracic radiology at the time, and that was very interesting to be part of that and seeing this whole thing unfold. And I decided to do a research project back then like everyone else on the Covid. And then there I got to know this guy called Michael Anderson, who is a thoracic radiologist at my hospital, and we did this Covid project together in my residency. And then he asked me if I wanted to do this Ph.D. project afterward. So that was very lucky. My way into medicine has been like, I’ve been trying & trying a lot of different things. I started doing research like actually ten years ago now in cardiology, and for a long time I thought I wanted to be a cardiologist and I was working with echocardiography and doing research in that and also in the clinic as a medical student, I was very lucky also there to have some very great teachers and people to look up to.

[00:03:39] Louis Plesner 

So I was on that path and then I decided I wanted to try some different things before, choosing a speciality. And I was in internal medicine when I graduated also in psychiatry because I had to really. But that was also incredibly interesting. And then I found that both in cardiology and in internal medicine and also in psychiatry, but that was obviously not so much. I was very drawn to medical imaging. So the ultrasound part in cardiology, but also chest radiographs and CT and the whole imaging thing, MRI of the heart led me to think that maybe I should just be a radiologist instead. So that’s what I decided. And from then, I haven’t really looked back because that was just so incredibly interesting. I really liked the speciality. I really liked the way we work there. Also, the kind of generalist feeling that it has that you don’t really necessarily just focus on the heart. You get to know something about everything that is incredibly rewarding and feel like the speciality is just so rewarding because you, really play a large part in the patient’s journey through the system, really. When the patient takes a turn in the diagnosis pathway, usually radiology, most many times radiology has to do something with that. So now I am a couple of years into my Ph.D. And I am trying to get it done by next summer and then I want to continue my residency for four more years until I can become board certified in radiology. 

[00:05:13] Jen Callahan 

So the residency then will be for radiology? 

[00:05:16] Louis Plesner 

Yeah. 

[00:05:17] Jen Callahan 

And then obviously, as you’re talking about the heart and your interest in echocardiograms and things like that, will that some physicians, radiologists choose a certain path or studies that they specialize in? Would you be more so looking for the cardio like thoracic? 

[00:05:37] Louis Plesner 

Yeah, that’s an interesting question because in Denmark and I know that it’s also like that in other parts of the world, but it’s very different. In Denmark, at least the cardiologists, they do all the cardiac imaging themselves. We do some coronary angiograms, but echocardiography is only here with the cardiologists and also the MRIs. You have to give that up and then focus on all the other things. But obviously, for thoracic radiology, it’s something that interests me a lot. And I could see myself pursuing that as a subspecialty at some point. 

[00:06:10] Jen Callahan 

So if you have just to clarify for that, for my own sense, say if you have a patient who’s in within the hospital who was there for a heart issue and is being cared for by a cardiologist and say they order a chest x-ray or, you know, a CAT scan of the chest, it’s not so much a radiologist will be doing the reading on that. It’s the actual cardiologist who ordered it. 

[00:06:31] Louis Plesner 

No. It’s more like the focused heart examinations will be done by the cardiologist. But these more broad diagnoses, both in terms of chest radiographs, but also CTs of the chest. They will be done, absolutely by radiologists. It’s only the specialized heart examinations such as MRI of the heart and echocardiography. 

[00:06:51] Jen Callahan 

Okay, understood. Gotcha. 

[00:06:52] Louis Plesner 

So we still and we still also work together on, for example, the CT, the coronary artery CTs, because radiologists describe all the parts surrounding the heart. So it’s kind of a joint venture there. 

[00:07:05] Jen Callahan 

So you decided to move into radiology and then you started with machine learning. What was the transition point from being with radiology into machine learning and for those who are listening, machine learning basically is what we’re talking about artificial intelligence with computers helping. Am I correct? 

[00:07:25] Louis Plesner

Yeah, absolutely. 

[00:07:26] Jen Callahan 

So how did you, like piece those two together that you decided to transition, almost like you put your radiology residency on hold so that you could pursue this further?

[00:07:37] Louis Plesner 

Absolutely. Yep. AI in radiology is something everyone is talking about now. And also when I started this in 2020, everyone was also talking about it back then and it was said that we would be out of a job in a few years. That was kind of the word on the street, and it still is in some sense. Actually, we are just in Danish national media. There was some guy saying that radiologists are a dying race, so we are fighting that a little bit. As I said, I’ve done like ten years of clinical research and I’ve been working also with all sorts of statistics before, but I’ve never done any programming or especially machine learning programming. So that was completely new for that. What interested me at the beginning of this journey a lot was that Michael, my supervisor, I have two supervisors. They are both named Michael. They wanted to do a study that could assess the clinical impact of these AI tools, and I thought that was very interesting because at that time that was a little bit new. It was more like the research was not so far ahead just three years ago as it is now, it’s going incredibly fast. So it was mostly traditional, you could say, medical research where you want to try some sort of new development in the clinic on some patients and see how it goes and not so much the tool itself that interested me, which is the whole AI technology. And then I have through these years, obviously gotten to know some more about this, even though I’m far from an expert on the mechanics inside the models, but I’m still more and more finding it extremely interesting and have also started now to work on some small hobby projects myself. But my Ph.D.Research work is on the models that others have developed, which I’m testing in the clinic, like translational research, or you could call it. 

[00:09:27] Jen Callahan 

So is it almost like companies that have already developed technology like this that you’re helping them do the basically clinical trials? 

[00:09:35] Louis Plesner 

Exactly. And people are pushing this technology forward without any real solid evidence. And that is still happening. Now evidence is getting better and better, but still, it’s in the beginning. So, yeah, that’s an issue that we focused on to be able to know that these tools are safe before deploying it. 

[00:09:51] Jen Callahan 

So a question for you. In the past two weeks or so, I’ve actually spoken to two different companies that have developed two different types of artificial intelligence to aid in the reading of radiographs. One is for the actual images themselves that will pick up on subtleties or help with misreadings or something that might be missed. And then the other company was more so on the side of helping the radiologists complete their report. So, you know, the radiologists sitting there looking at the image, putting everything into the dictation of what they see. But that technology gives the impression, which is like the summarization of what they’ve seen within that is, or one of those areas where you’re more leaning towards or is there another area of AI that you know is out there that I haven’t even heard of yet? 

[00:10:44] Louis Plesner 

I think I can be put anywhere in the whole patient journey through the department really. And applications are coming from all corners. And what is the most sexy part or most interesting part is to make the diagnosis. But that may not be where I can have the biggest impact because radiologists are pretty good at that and it’s difficult to gain a few percentages and it’s also difficult to show it in the study that this actually does something. There are also all sorts of areas surrounding the for example, looking in the patient or reading the image request form and assigning it to a protocol or also billing and all sorts of stuff. From my standpoint, it’s more obvious that it can help us a lot in these areas because also that’s some of the things that we don’t really want to do. We like to give the correct diagnosis to the patient, but not so much doing all the protocols all the time. Yeah, that is also something actually we are working on in our group with some of the new large language models such as ChatGPT and others that are booming since last year. 

[00:11:51] Jen Callahan 

So let’s talk a little bit more about you and what you’re currently doing. So you have your Ph.D. thing going on the side, but then as I said, you currently work part-time within another hospital, doing part-time work within the department and you had said that you’re assisting or teaching really with the ultrasound aspect there. Do you want to give us a little bit more information? 

[00:12:12] Louis Plesner 

Yeah, that’s something completely different from what I do in my Ph.D. I enjoy very much to have patient contact, even though I have when going into radiology, you can still have a lot of patient contact in radiology as well, doing interventions and ultrasounds. And so I wanted to have some patient interaction and also keep in touch with, kind of my identity as a doctor while doing this, delving into this world of machine learning, and then I saw this, that they were looking for someone to kind of introduce ultrasound in the emergency department in another hospital in Denmark. At least that is quite new. Not so much new, but maybe the last. Maybe ten years or so. It has been growing like the ultrasound competencies of non-radiologist physicians. And it’s still like in Denmark in the absolute beginning. But in my opinion, ultrasound point of care, ultrasound, it will grow more and more in the clinical specialties in all specialties, really, because it’s just so incredibly useful and it’s right at hand with and can many times make a big impact compared to the stethoscope or it’s more much more old fashioned. So I took this job where my role is to kind of work as an emergency physician, seeing patients and treating patients, but also having this particular focus of teaching everyone else point of care ultrasound. And so I’m helping them with ultrasound interventions, pleural drainages and also placing a line with ultrasound, and then I’m teaching them how to look at the liver or the gallbladder or the lungs or the heart, because I, from my previous experience, have quite a lot of experience in echocardiography. So yeah, that’s kind of a teaching physician role that I’m doing as well, which I really enjoy a lot. 

[00:14:10] Jen Callahan 

So the machine learning and doing your Ph.D. In that, do you actually take classes for something like this or is it basically like you working with Michael or somebody, someone else within a school setting doing this research?

[00:14:24] Louis Plesner 

Yeah. So we, have contacts where various contacts much more specialized in this than us in data scientist guys that we have worked with. But like I said, the main studies I’ve been doing in my Ph.D. is more like clinical validation or clinical trial and not so much development of these models. So for that, we are actually doing it ourselves with the help of biostatisticians statisticians, and just like much research in medicine really. And then the. Working on the models that we are doing ourselves as doctors in the group is considered at this point much more like a hobby. We have been working, for example, with natural language processing ourselves and doing some preliminary models that can read X-ray reports. But now this is all obsolete due to ChatGPT. So which is just excellent at all this language stuff. Radiology is just a small, tiny part of what it can be used for. 

[00:15:26] Jen Callahan 

For radiology though, let’s go back to that to like some simple, fun questions, obviously, you said that you love cardiology and stuff. Is there specific exams that you like looking at besides chest radiographs or something? Do you like MRI or do you like CAT scan more or do you lean more towards ultrasound because that’s what you do in your part time thing. 

[00:15:48] Louis Plesner

In my first year as a radiologist, I haven’t been working a lot with MRI. It sits more with the more experienced radiologists, a little bit MRI of the brain I have been looking at. But other than that, the musculoskeletal MRI is something that I have not really delved into so much. So what I have done most is plain radiographs, ultrasound, and CAT scans. I enjoy all that. I really can’t say if there’s something I enjoy more. Sometimes you get through a lot of examinations that really feel like doesn’t really impact so much the patient, and then you get these examinations where you may make a lot of impact, and that’s basically even the one modality or another modality. When that happens, it’s just always fun. But if I was to say something, I would probably say ultrasound. 

[00:16:34] Jen Callahan 

Ultrasound is obviously useful. There are all different modalities. They’re useful, but it’s great because it was very versatile in the fact of one, that there’s no ionizing radiation that’s being used. But then I currently work in interventional radiology and like you said, to look at vessels and stuff, but then you can also see different body parts and I agree with you there. 

[00:16:53] Louis Plesner 

And having the patient interaction as well, just makes it a little bit more fun if you could have it with the patient interaction while going through the. That would also be fun if you like the patient interaction like I do because you feel like now the patient has reached the right place where someone can actually make a diagnosis or say something that feels very good. But then also you’re very poor where you can really not see anything due to bowel air or something, and you just feel like, I want to be a gardener instead or something. 

[00:17:27] Jen Callahan 

A gardener. So the school that you’ve been working with, Herlev. You had said? 

[00:17:31] Louis Plesner 

Yeah, the hospital? Yeah, That’s a hospital. Yeah. 

[00:17:34] Jen Callahan 

Do you think that they’d be interested in developing or are you working hand in hand with the Michael that you had started with, maybe somewhere in the future? Is that something like a dream of yours, maybe to develop something with machine learning? 

[00:17:49] Louis Plesner 

I wouldn’t say that is my dream. No, that would be something I would be interested in as well. I have a lot of interest, but I think that also depends if you want to do it in a research context to make a small project or if you want to develop a product that can be sold that can help patients across the world because that’s a completely different story and a completely different job, which is a startup basically, and not something that I would consider myself particularly interested in. So if you want to make an AI product that can be sold across the world and help patients, then it’s a completely different story from doing a research project because then you’re basically doing a startup and it’s another kind of job than being a doctor, even though it must be very rewarding as well and interesting, I would consider myself more of being a traditional researcher and also keeping up with other research interests that we have. Also working with advanced CT scans and other things we’re doing. I just want to take this technology and see what good we can make with it, both for the radiologists and for the patients. That’s also with the project that came out in radiology. The research project that led me to this podcast, it’s about actually automating, doing autonomous reporting with AI of chest radiographs and something that, that we are interested in from the perspective of being able to alleviate some of the workloads in the radiology departments. So from more of a standpoint of what good can we do here. 

[00:19:17] Jen Callahan 

I feel like we’re coming to an end here. Is there any words that you want to put out there to the radiology world for the future of AI or just for radiology in general? 

[00:19:27] Louis Plesner 

Yeah, like I said in the beginning, I want to say to all the junior radiologists like myself that don’t be afraid to go into this extremely cool and nice specialty just because some people are saying that we will be out of a job, I don’t think we will. You can never predict the future, but if radiologists will be out of a job, I would say that all other physicians will be out of a job as well. The algorithms would have to be that advanced to put all the physicians out of jobs and that could happen. But you cannot really think in that way because then we would probably just do something else. That would be my statement.

[00:20:01] Jen Callahan 

I agree with your statement. I definitely agree that radiologists are not a dying breed at all, and I don’t think doctors are a dying breed. No, people are needed to care for people. They’re great words of advice and I wish you luck with the continuation of your research into machine learning. Louis, thank you so much for taking time with us, everybody. This was Dr. Louis Plesner with us, joining from Copenhagen, Denmark, and hoping to hear from him soon again. You know where he’s at in his journey through education and radiology. So thank you so much, Louis. 

[00:20:33] Louis Plesner 

Thank you so much for asking me. 

[00:20:35] Jen Callahan 

All right. We’ll see you next time, guys. You’ve been listening to the Skeleton Crew brought to you by xraytech.org. Join us on the next episode to explore the present and the future of the rad tech career and the field of radiology.