David Larson Official Transcript

Intro/Outro – 00:00:02:

Welcome to Frame by Frame: Rethink Imaging, a podcast by Imologix. Here, we explore the intricate world of medical imaging, aiming to dissect the field and inspire both professionals and curious minds alike. I’m your host, Chris St. John.

Chris St. John – 00:00:19:

Today on Frame by Frame: Rethink Imaging, we are excited to welcome Dr. David Larson, a prominent leader in the field of radiology and a dedicated advocate for improving healthcare delivery. Dr. Larson served as the acting chair and executive vice chair of the Department of Radiology at Stanford University School of Medicine, where he has played a pivotal role for 11 years, including serving as vice chair of education and clinical operations, as well as the associate chair of a performance improvement. Dr. Larson’s work is characterized by his focus on integrating complex adaptive sociotechnical systems theory into healthcare, enabling innovative and sustainable solutions to the intricate challenges facing the field of radiology. A seasoned pediatric radiologist, Dr. Larson previously held significant roles at the Cincinnati Children’s Hospital Medical Center and Utah Radiology Associates, further enriching his expertise in the domain. His dedication to Dose optimization, patient safety, and the effective use of technology in radiology has positioned him at the forefront of the discipline. Throughout his career, Dr. Larson has been recognized for his contributions to medical imaging education and performance improvement initiatives, making substantial impacts on both clinical practice and patient care. Join us as we start our conversation with Dr. Larson and his insights on Dose optimization and exploring the future of radiology through his perspective. Thank you for putting up with that introduction, Dr. Larson.

David Larson- 00:01:45:

Thank you so much, Chris. It’s a pleasure to be here.

Chris St. John – 00:01:46:

It’s awesome to have you here. I always feel so funny with the long intros, but I think it’s important that people hear your background and learn a little bit about you before hearing what you have to say. Now, just like off the bat, after reading your introduction, we took this verbiage from your LinkedIn, but integrating complex adaptive sociotechnical systems and integrating them into healthcare, what does that mean?

David Larson- 00:02:11:

That’s a mouthful. Yeah. Well, so my background is in quality and performance improvement. And so most people have a conception that quality and quality improvement is mainly about just making sure that we do the right thing most of the time and that most of the time we do. And it’s just about making sure that we actually do. In reality, in order to make things go right every time or most every time, in order to make things continuously improve, it actually takes a lot more than that. You actually have to be able to change your organization, continuously change it in a way that you change the technology and integration of the technology into the workflows, change the organizational structure, change how people are held accountable for their work, how they’re coached and supported. All of that throughout the organization goes into making sure the organization performs well. So a complex adaptive system, and especially a socio-technical system, means the social part and the technical part. So you’ve got to have the technology in place and the analytics and the dashboards and so forth. But you’ve also got the social part of any organization, and that’s the interrelationships between people, how their reporting structure is in place, how they are trained and supported, even how they feel. So it’s a wide spectrum from the social side to the technical side to make this all work together. So that’s my background. I’m very interested in the whole spectrum. What we’re going to talk about today is much more on the technical side, but I’ve also had the pleasure of working much more on the social side around things like radiology peer review, converting that to peer learning, how radiologists, work with each other and learn from each other. But the whole thing all comes together if you want to make the thing work really well consistently.

Chris St. John – 00:03:51:

Yeah. Can you just expand a little bit? You said peer review to peer learning. Can you talk a little bit more about that?

David Larson- 00:03:58:

Sure. Yeah, good. So that’s the other side of the sociotechnical system that we’re going to talk about today. So I’ve had the pleasure to work in a variety of roles, including my role at Cincinnati Children’s Hospital. When I walked in the door, one of the first things they said to me is, congratulations, you’re brand new, pretty much straight out of fellowship, one year out of fellowship. And now you’re in charge of peer review. And peer review means in radiology generally that you radiologists score each other on the cases that have come in and basically find problems and give that feedback into a system, submit those problem cases where there is a misdiagnosis. And then those problem cases are handled and no one really knows what happened is it’s not really well defined as to what happens after that. But there is presumably measurement and then you can take the outliers and move them out. And so you can imagine the social aspect of that right? Peer review that has never caught on really because it’s problematic in many ways. So one of the first things I did early on in my career other colleagues from around the country is that we converted that social paradigm from reviewing each other’s performance and scoring to finding opportunities where performance can be improved and that may be a miss or a great call but either way we actively look for those opportunities in a constructive positive way that contributes to a better culture, a culture of learning and support instead of a kind of gotcha culture.

Chris St. John – 00:05:24:

Yeah, for sure. I feel like lots of industries are honestly taking that approach these days, right? Like we’re making this shift culturally just to ones more of understanding, honestly, more than anything.

David Larson- 00:05:38:

Yeah, and I think that that’s what’s interesting about the concept of the socio-technical system, these complex socio-technical systems, is you’ve got to have the heart as well as the technology, right? And so as we’re going to talk about CT radiation dose optimization, most of that is on the technical side, but there’s still also the professional and interpersonal part of it.

Chris St. John – 00:05:57:

Right. So this is now the second episode I’m doing on Dose. I was just talking to Dr. Don Frush over at Duke last week. And we were talking a lot about like dose safety. We dipped our toes into some cumulative Dose stuff. You know, we talked all over the place, but I feel like as I’m trying to learn about everything going on in radiology, I feel like there’s so much emphasis these days. I’m just hearing Dose, Dose, Dose, Dose, Dose. For better or for worse, I’m not making a qualitative statement about hearing about dose, but just for my own purposes, let’s start at the beginning again. How would you just, in basic terms, explain radiation dose and why the optimization of Dose is so important?

David Larson- 00:06:37:

Sure. Well, radiology, when it comes to imaging modalities, there are a number of ways that we can see inside the body, right? So MRI uses magnetism, ultrasound uses sound waves, and x-ray and CT use x-ray radiation. And an x-ray is basically you point radiation at a person and it creates a shadow on a detector. And a CT, you do the same thing, but you do it in three dimensions as it goes around the patient. So radiation is necessary in order to acquire those images. But the challenge is that, well, I’ll say, first of all, a CT is miraculous. I mean, it has transformed medical care. Really. It is. What we’re able to see with CT is phenomenal. So we wouldn’t want to ever downplay the transformational nature of the technology. But at the same time, it does use radiation and radiation is associated with increased risk of cancer. It uses low dose radiation, but we want to minimize that risk in any way we can. So there is a core principle in radiology, as low as reasonably achievable. We use dose as low as we can. We need to use some dose, but we don’t want to use any more than is necessary. The problem with CT is that the more dose you use, the better the images look. You know, the thing that is your limit is pretty visible to everybody. If you’re not using enough radiation, then it kind of jumps out at you. Say, hey, I can’t, you know, it’s very noisy is how it looks grainy kind of images. Like you’re not gonna be able to make the diagnosis. On the other end of the spectrum, there’s nothing that jumps out at you that says, hey, you’re using too much dose. Right. So it’s kind of like if there’s a threshold and the threshold isn’t exact point, but there is a threshold below which if you go, it’s going to be too noisy. But if you go above it, it’s not necessarily a problem that you can see. So then the goal is to kind of get to hug that lower threshold as closely as you can. But it’s tough because you don’t know where that threshold is innately unless you’ve gone over it. Right. So that’s one problem or it’s one challenge. And the other challenge is it’s very, very sensitive to size because as you create a shadow on the detector, some of those x-rays are absorbed. And if you have greater size, more of those x-rays have to be absorbed in order to create that shadow. So it involves a lot of math and involves a lot of thorough understanding of how the machine actually works at every size. And that understanding is not really well, I’d say, developed and well implemented, at least not to the level of sophistication that we really need to fully optimized dose.

Chris St. John – 00:09:07:

Right. And so is that so you’re saying, obviously, like I hear the term size adjusted dose being thrown around as well. It’s my assumption. Is that some sort of automatic calculation? Is that something that people are doing? Like, how do you end up with a size adjusted dose?

David Larson- 00:09:22:

Yeah. So one thing I’ll say about optimization before I address that specifically, it’s important to remember that optimization of dose is based on image quality, right? So it’s not actually about the dose that the organs receive. I mean, you want to minimize the dose that the organs receive, but as long as you get adequate image quality and use the most dose efficient parameters on your scanner, then you have to assume that’s the lowest dose your organs can receive. That image quality ends up being the driver, right? So the amount of absorption that absorbed by an x-ray as it passes through the tissues, it’s about one inch where half of the x-ray has been absorbed, if you think about that, right? So if you think about most people on average are about maybe eight to 10 inches in diameter, if you make it an average diameter, you’ve now absorbed, it’s every inch, it’s absorbed half of the dose. And by the time it gets to the output, you know, it’s down to a small percent on the input, right? On the entry level versus the exit level, right? So every time you add just another inch, then you need to increase the dose in order to keep that amount that’s coming on the exit side, roughly similar. Does that make sense?

Chris St. John – 00:10:35:

Yeah. So every inch is a half of a half of a half of a half. I’m not a math guy, but that’s not exponential. Is it exponential?

David Larson- 00:10:41:

Yeah. Yeah. It is.

Chris St. John – 00:10:43:

Okay.

David Larson- 00:10:43:

You got it. Cool.

Chris St. John – 00:10:45:

See, look at me. Look at me. I’m on fire today.

David Larson- 00:10:48:

That’s right.

Chris St. John – 00:10:48:

So in terms of like implementation, what are the basic steps that radiologists are taking to ensure that they’re hitting this magic window? How do you go about critically thinking about the approach to hitting it? You know?

David Larson- 00:11:03:

Yeah, it’s a great question. So the general approach that’s currently taken right now is that you kind of do your best to hit that window kind of a, I think you described it well, right? Basically you look to see what are other sites able to hit and just make sure that your doses are in that range. And if they’re not in that range, then you make adjustments until they get in that range. And what I’ve found that most sites, if you do that, you know, reasonably well, you certainly will improve over where you were, if you didn’t have any strategy in place, but you still get to the point where there’s enough variation that on a given patient of a given size for a given technique, the patient who receives the highest Dose who’s of that size compared to the patient who receives the lowest Dose, it’s about a factor of two, about a factor of a double. That means that there’s a significant amount of improvement to go with you now set a specific target and consistently hit that target as a function of patient size. So that is not really in place yet, right? That’s the goal. That’s really what dose optimization is about. And that’s that next step to go from just kind of managing to make sure they’re roughly within the range to saying, no, no, we’re going to nail it every time. We’re not going to tolerate double. We’re going to like nail that every time.

Chris St. John – 00:12:17:

Yeah. And so how do you go about optimizing then?

David Larson- 00:12:19:

Yeah. So really what it comes down to is set a target and consistently hit that target, right? The shift in mindset. And this happened in manufacturing really about, and it’s now about 60 to 70 years ago, where across the world, you know, this was the kind of lean six sigma quality transformation, you know, really kind of focused in Japan where the mindset used to be, you know, you just keep, you have specification limits. And as long as you’re somewhere within the limits, then you’re fine. It’s been termed the goalpost mentality. And so this approach is now a target based approach. And so that’s what true optimization is, is you say, I want to specifically hit this target and I’m going to continuously work to make it so I am very precisely hitting that target. And so to do that, you have to start with the target. In our research, we started, we use estimated image noise as a function of patient size, but that’s a bit more complicated and probably more than is necessary. So we’re now looking at using what’s called a CTDI vol, CT dose index for a volume, right? The CT dose index. So if you can set a target for your CT dose index over a range of patient sizes, you start there. Then it’s a matter of setting the parameters on your scanner to hit that and then have a monitoring application to make sure that you do so on a regular basis.

Chris St. John – 00:13:36:

Right, right. And so you’re saying that like the dose monitoring stuff is adjusting dose in real time. Is that correct?

David Larson- 00:13:46:

Kind of. So I think what you’re referring to is called automated tube current modulation.

Chris St. John – 00:13:53:

Sure. I was going to say automatic tube current modulation, but you know. Exactly.

David Larson- 00:13:58:

You did. Yeah. That’s what you described. Yeah. Tube current is basically a proxy for the amount of radiation that’s going out. And so from the x-ray tube, you put a certain amount of tube current in and you get a proportionate amount of radiation out. So it’s kind of like you could saying it’s like automatic radiation modulation. Right. And modulation just means it adjusts as it goes around. So tube current modulation is a technology that’s been around for about 20 years. And as it goes around the patient, it recognizes how far it has to penetrate, how thick the patient is at each angle. And so whether if the gantry is front to back, then it’s usually thinner. And if it’s side to side, then it’s usually larger. Right. So it automatically adjusts. It’s radiation as it goes around. And it’s a great technology. It works really well. The main downside to it is that even though it adjusts, it doesn’t clarify what it needs to adjust to. So it adjusts. And so what we tend to see is there’s still a lot of variation, still used too much radiation, more than was necessary to hit that target Dose or that target image quality. And so you have to really nail down what is that target image quality. And then on the backside, confirm. That you hit that target. And that’s what a tube current modulation algorithm does not necessarily do for you. That’s why we’re still facing this problem. Yes, it adjusts as it goes around. So it’s, you know, it’s a great technology. It’s just they didn’t quite finish it. It has to be accompanied by a monitoring system. And that monitoring, you have to be fully confident that you’ve nailed that target Dose based on image quality.

Chris St. John – 00:15:35:

Right. Maybe this is a stupid question, but it’s the first thing that comes to mind. Like I know about phantoms when setting up your machine and stuff, like has the radiology community been using cadavers, phantoms, like what are the different methodologies of getting better at this? We kind of doing it all?

David Larson- 00:15:53:

Well, a lot of it can be done by phantoms. I’d say the real question is, it’s not exactly clear what is the amount of image noise that can be tolerated on a given study? You know, what is that exact threshold? And, you know, the methods to get to that are really right now, it’s still focused primarily on what the radiologist thinks looks good, right? What they think, you know, is reasonable and methods to try to advance on that actually haven’t really found that there’s much better than just actually having the radiologist look at it and say, yeah, looks pretty good. So, but that’s hard to quantify and that’s really what’s needed. So there are sites that are increasingly doing that. That’s what we did at Cincinnati Children’s. We actually gave a score to it. We gave a number and others have done that as well. What’s needed then is a universal number, some type of acceptability standard to say whether it’s acceptable and what target curve does that translate into? And then do you consistently hit that every time?

Chris St. John – 00:16:48:

Yeah. And so it’s not just the radiologists who have a voice in this conversation, right? It’s a collaboration between physicists and technologists and IT, right?

David Larson- 00:17:00:

Yeah, for sure. Yeah.

Chris St. John – 00:17:02:

I’m just curious, like what that collaborative process actually looks like, right?

David Larson- 00:17:06:

Good. Yeah. So, okay, good. So here’s the role. So you’re right, especially when it’s done well, it’s that collaboration between, just as you said, the radiologist, the physicist, and the technologist. So the radiologist’s job, and by the way, they may overlap in terms of their skills. So what they bring to the table may vary based on a local practice. But in general, you know, when you’re functioning in the role of radiologist, your main job is to say at what point the image quality is acceptable versus it’s too low. And if you can give some indication of it’s too high, that’s helpful, but that’s often harder to do. So then the role of the radiologist and the physicist working together is to say, okay, what are the parameters used on the scanners that will get us to that target curve, right? So we establish a target curve based on what the radiologist deem acceptable, you know, how much noise can be tolerated. Then the radiologist and the physicist work together to figure that out. And then you come up with protocols. And then the protocols have to be deployed, often a very large fleet of scanners. Often they’re different manufacturers for different reasons. They have different behavior. And that’s complicated, right? So the technologists have to really work hard to make sure that that’s all in alignment. And then there needs to be a monitoring system to make sure that it’s all happening on a regular basis and report it up. And this is where I think there’s probably the biggest deficit. Report it up on a regular basis to the management of that practice to make sure that it’s happening consistently. That’s almost completely absent, I would say.

Chris St. John – 00:18:30:

Okay, I want to go down this protocol rabbit hole now. You already said there’s a bunch of people involved. There’s a bunch of equipment involved. There’s software involved. Like, how do you approach the development of protocols?

David Larson- 00:18:42:

Yeah, great question. If it’s not reduced in complexity, the complexity can overwhelm you very quickly. So you got to take it one step at a time. So there are certain principles that are just kind of well accepted. For a certain size, you increase your tube voltage, your KV is what is called a KVP. You increase that with certain thresholds in terms of like if you once you get past a certain size, you move from 80 to 100 to 120 to 140, depending on your scanner type. So that’s one variable. Another variable is the gantry speed and the pitch. So that the gantry speed means how fast it goes around the pitch means how much overlap there is between the slices. For each of those, you basically set a parameter that is appropriate, and then you adjust the other factors accordingly. So once you get those set, then it mainly comes down to your tube current modulation algorithm, what we talked about before. And that is the thing that if you set that to a certain amount based on size, that’s the thing that it’s going to adjust and result in your dose, and you want it to be as low as possible based on your target quality. So it really comes down to figuring out that tube current modulation algorithm parameter, how you set that tube current modulation relative to the outcome, if you can control all the other variables, I would say there’s some additional complexity that comes in, where there’s a feeling as you can, there’s so many use cases, that you set different, you know, image quality standards based on every single use case. And what we found is you can kind of reduce that down to about three, four use cases, you know, it’s basically your routine, like your routine CT abdomen and pelvis, for example, if you need a higher Dose for things like trauma or CT angiography, you set a certain target level for that lower dose for things like lung windows, and then maybe one more for things like, you know, bone windows for routine studies. So it really, you can reduce the complexity down. And then you can, once you have those few curves, you have just a few target curves, then you can now expand those to all of your protocols and monitor your protocols one by one. And then you do that for each scanner type. So you reduce a, what was an overwhelmingly complex problem to a still difficult, challenging problem, but one that’s manageable.

Chris St. John – 00:20:56:

I was going to say, it still sounds difficult and confusing to me, but it does make more sense now, at least.

David Larson- 00:21:02:

Yeah. And so that’s where you have to break it down one by one. And that’s where you really have to monitor it. And so the other thing is this takes people, right? It takes experts, it takes systems, it takes the IT systems on both the monitoring side and the management side. And so to do that, that takes resources. And the way we’re paid in radiology is per scan. And so you’re not directly paid to do this. Now, the payers would argue, and I think fairly so, is in the per scan payment, but a local institution isn’t directly paid for that. So then they have to decide how many resources, how much they’re going to spend on a CT dose management program. And when you have very tight budgets and you’re already relying on radiology, which produces a lot of revenue to subsidize the rest of your hospital, it makes it tough to build that whole program. So by default, then what ends up happening more often is you do your best with what you have and you end up with what I described, that ends up being that, yeah, you’re roughly within this range. And it’s about a double of 2x difference between the highest Dose and the lowest Dose that we give in patient size.

Chris St. John – 00:22:07:

So, I mean, you’re talking a lot about like financial implications, time implications. And I know there’s a massive staffing shortage going on right now. It comes up every single episode. And part of the show is wanting to get the word out to people that like there’s a career path here. But what would you recommend to folks at smaller enterprises, at smaller hospitals with less financial resources, with less staff members in terms of trying to optimize Dose and be most efficacious with their time and with their funds? Like, what would you recommend to smaller places struggling with this?

David Larson- 00:22:40:

Yeah, well, a couple of things. It’s a great question and one that we grapple with as a entire profession. So the first thing I would say is one is kind of the mentality of dose optimization and to just adjust the mentality from, you know, getting it roughly within certain parameters to say, set a target and consistently hit that target. Once you adopt that mentality, then it will drive you to, first of all, start monitoring your actual doses and to have a target. Right. And once you have that, then it can be, you know, one step at a time to figure out what is that target and then, you know, adjust your protocols on a regular basis. It’s surprising how much you can do just with some regular adjustments to your protocols to make sure that they get closer to that target. So that’s the first thing. And the practice of any size can do that. It’s not super easy, but it’s totally doable. And then I would say what we’re working on through the American College of Radiology, one of the other hats that I wear is on the chair of the Quality and Safety Commission there. And we’re setting up a group sites through our learning network that are working on this problem specifically and are discovering techniques and sharing their techniques. And we hope to have a number of strategies that can become more widely available or bring sites into the program and will help you get to that point as we’re developing this capability.

Chris St. John – 00:23:56:

That’s amazing. Is that going to be like a program folks will be able to enroll in?

David Larson- 00:23:59:

Yeah, it’s actually available now. So the ACR Learning Network currently has four what we call improvement collaboratives. So there are four areas of focus that we try to help local sites, actually the sites help each other, improve their performance. And so two are based on image quality. One is in mammography and one is in prostate MRI. And then the other two are based on one is on following up on recommendations for incidental findings, in this case, lung nodules. The other is around improving your lung cancer screening programs. And then there’s a fifth one that is whatever problem you have, bring it on in to the program. And we’ll get other sites together that are either working on the same problem or something else. And we’ll teach you the methodology to go solve whatever problem that you’re working on and help support you in that. So yeah, this is currently available. And we have a number of pioneer sites with CT Dose optimization. We hope that we’ll have increased learning and can help local sites who want to solve this specific problem. And come join us and we’ll help make that happen.

Chris St. John – 00:24:58:

That’s really cool. You’re training folks, which is awesome. What educational strategies have you found most efficacious when bringing folks up to speed on all these optimization techniques?

David Larson- 00:25:10:

Yeah, it’s a great question. So as we started our conversation with, you know, kind of talking about the socio-technical system, some problems in radiology are really very skill-based, for example, and require everybody knowing exactly how to do it really well. And you find that with x-ray, you find that with mammography, ultrasound, and I’d say interpretation skills are really primarily around skill. Others are much more technology dependent. And I would say this is one of those. CT radiation dose optimization. Start with getting a hold of your protocols, right? You’re setting that target, hitting that target, figuring out your tube current modulation algorithm, and so forth. That starts with a small team. So intense education or training or learning, I guess is probably a better way of saying it. They need to go figure that out exactly how to do that with a small team. And once you can get your protocols under control, then you can implement those protocols. And then it’s a matter of educating, training, monitoring your whole practice, right? So then it’s things like, well, to make sure the patient is positioned correctly and the arms are, you know, up and, you know, it’s using the right protocol in the right situation. That needs to be done broadly. But where I see sites get hung up is they start with that. And they haven’t gone to the thing that’s the most important, the greatest source of underlying variability, and that’s those protocols. So the education, training and learning starts with a small group. Once that’s really under control, and has the technology to support it, then it rolls out to the larger practice.

Chris St. John – 00:26:33:

Okay. This is one of those moments where I don’t have a follow-up locked and loaded again. Should we go more into the technical? Is that like something that you feel is worth diving into a little bit more?

David Larson- 00:26:44:

We can. It generally only applies to the people who are going to dive into that. I’m happy to do that.

Chris St. John – 00:26:49:

It’s totally fine. I’m just feeling out the next phase as I think about where to take a turn to.

David Larson- 00:26:54:

Yeah, there’s also kind of the layers of oversight and responsibility of the organization. So we could talk about that.

Chris St. John – 00:27:01:

Can you talk a little bit about the layers of responsibility in an organization and how they apply to Dose optimization?

David Larson- 00:27:09:

Sure. Great question. At the basic level, we generally tend to focus on the very specific techniques that need to be applied and the people who do that, who apply the application of those interventions or changes that need to be made. So at the lowest level, it’s determining what is that target curve for optimization, it’s then determining how you get there and monitoring. And that needs to have a small team to make that happen. And then the next layer is, okay, once that’s embedded into your protocols, how you monitor that and make sure that it’s all happening. What is often neglected is the layer above that, or maybe two layers above that, that are required to make this whole thing work, right? So somebody at some point has to say, we’re going to allow people to have the time, we’re going to write in their job descriptions, and that’s going to be their job, especially in a time of constrained resources. That’s only a manager, right? Or director or whatever title you have to give them. It’s somebody in charge. Somebody has authority over the resources and the time and can hold people accountable and quite frankly, hold themselves accountable. They also then need to have a reasonable understanding of optimization of CT dose optimization. And again, I would say the same thing to them, as I said earlier, to anybody really comes down to setting a target and hitting a target. So what those managers… Those supervisors or leaders need to do is to say, okay, we’re going to commit doing the program, commit to the resources, get the tools in place, including a monitoring system, have the people, dedicate their time, have job descriptions, make those investments. And then what they should expect is to see the data, right? And now they can’t see all the data because it’s overwhelming, right? But they need to see, to verifiably see for the doses, they are setting a target and they’re hitting a target consistently. If they can see that. If they just draw that target out, size on basically patient size in terms of water equivalent diameter on the X axis and CTDI on the Y axis, and then plot all the points. And if it’s a big scatter, then it’s a problem. And if it’s super tight, right around a curve, then it’s great. That’s what they can expect. They should expect from their investment. And then I would say, then there’s a broader layer and that is the regulators and the payers. And that’s what they should also expect, honestly. And they should have mechanisms to do that, to say, okay, if, you know, in terms of Dose optimization, we would like to see your graph, that graph I just described and see how tight that variation is. If it’s all over the place, then you’ve got some work to do. If it’s super tight, then you’ve clearly done that work.

Chris St. John – 00:29:46:

As departments are taking actionable steps to get themselves onto that curve, what do you think the most critical priorities are when trying to optimize?

David Larson- 00:29:56:

It really comes down to that setting the target and hitting the target and making sure all your parameters are there to make that happen. And it’s the mindset. It’s the target-based mindset rather than the goalpost mindset. And I know that it sounds repetitive, but that’s really critical because as soon as you convert mentally and your whole program converts to that, then you now start to focus and now you can improve. That’s where continuous quality improvement kicks in. And if you don’t do that, if you think that we just hit the goalposts, then we will always in perpetuity, I can guarantee you, you will be overdosing your patients until you adopt that mindset. That’s the mindset shift that’s required to not overdose your patients.

Chris St. John – 00:30:37:

Yeah, I’m not worried about repetition. I was just talking to someone earlier this morning, and it’s like, if two plus two is four and eight divided by two is four, then yeah, four is the answer to both questions. That doesn’t mean that we’re not answering different questions just because we ended up at the same place, right? And so if this target mindset is the move to get all of these different things like onto the same page, then that’s a damn good answer.

David Larson- 00:31:00:

Yeah, it really is. And so I’d say there are other things that are important. I don’t want to downplay those. So, you know, there’s equipment, for example, that, you know, modern equipment in just the past decade or two, like there have been continuous advancements in detector efficiency, for example. And now with a new paradigm with photon counting CT, you know, additional advancements, right? So I don’t want to downplay that. It’s important to have your equipment updated. And then your reconstruction algorithms, there have been advances there. So about a decade ago, there was this concept called iterative reconstruction.

Chris St. John – 00:31:33:

Yeah, I’ve heard about this, right? There’s iterative construction. I was reading about iterative construction. I was reading about automated exposure control. Is that, that’s unrelated though.

David Larson- 00:31:42:

That’s tube current modulation.

Chris St. John – 00:31:44:

Oh, that is. Okay. Okay, cool.

David Larson- 00:31:45:

Yeah, automated exposure control. So iterative reconstruction just is a method, a reconstruction method. So reconstruction means that you take the input data from, you know, that the scanner went around and you have a tube, an x-ray tube on one side and the detector on another. And as it’s going around, it’s generating just tons of data, right? Tons of information. And you can process that data and you have to process the data in a way that gives you an image that looks like a human being, right? So, you know, historically what they’ve relied on this concept, this approach called filtered back projection. That was the first generation. And then the next generation was called iterative reconstruction. And so it used computer techniques that were more computational heavy, but generated better images and meant you could use lower Dose to get the same image quality. Now there are AI based reconstruction techniques. They go even further beyond that iterative reconstruction that can get you even to a lower dose. So those are important too. So those are actually kind of the three pillars. I would say CT radiation dose optimization. It’s having your equipment up to date and then, you know, using reconstruction techniques that are the best ones that will give you the lowest Dose. And then it’s having a good process control system in place that is focused on optimization. Those three things. And the first two are the easiest in a way, like buy the equipment. You know, financially, it may not be the easiest, but it’s, you know, it’s straightforward. And then use the reconstruction techniques, like purchase the reconstruction techniques that will get you there. That’s the second. That’s also straightforward. Having an optimization and process control system is not so straightforward, but it doesn’t necessarily cost you any more money. It just requires a way that you manage your protocols. It’s the approach to doing it.

Chris St. John – 00:33:27:

As all of this new technology is coming out and as the field is evolving, like what would you say to those entering the field right now in this like period of massive change?

David Larson- 00:33:40:

Yeah, well, so there’s some things that are changing, but some things are kind of very similar to what we’ve seen before. So AI is what we constantly think of in terms of change. Well, AI in terms of this specific use case, AI reconstruction is just another reconstruction technique building on iterative reconstruction. Iterative reconstruction came out right in the middle of when I was doing all this research before. And so we had set those targets. That’s about a decade ago, about 12, 13 years ago. That’s when these came out. It’s very exciting, actually, because it allowed us to about cut our doses in half. So we’d set all of our tube current, our target curves based on filtered back projection. And then all of a sudden, iterative reconstruction came out. It’s like, oh, no, what are we going to do now? We have to redo all that work. And the answer is actually no, because, again, we set it based on subjective image quality assessments by the radiologist. In other words, the radiologist look at it and see how does it look? And it was a reasonable way to determine what are those target curves. And we just adjusted the target curves. It’s probably about 60 percent of the Dose that we had before. And then with AI reconstruction, we’re kind of looking at about the same stepwise function looking about another, you know, cutting in half or a little more than half, maybe about 60 percent of dose compared to what we had before. So that’s one thing that’s changing. And I’d say, don’t panic. You know, it’s actually not changing that much. And it’s very doable. And I would say the same thing with photon counting CT. It’s just coming out and we’re just getting familiar with it. But it’s still a similar approach. It’s optimization. It’s more about the shift into the optimization mindset. The other thing that we’re seeing changing is increasing regulations coming out, at least currently in terms of reporting requirements. So I would say it’s probably not exactly the way I probably would have framed it if I were to put those regulations in place. But that’s OK. You know, it’s mainly about reporting. I think the main thing to understand is that just reporting your Dose does not improve your Dose. Right. So you’re not done.

Chris St. John – 00:35:35:

Yeah, if I get straight Fs, I know that I’m failing, but I am not necessarily improving my grade.

David Larson- 00:35:40:

Yeah, and I would say it’s not even clear like what is an A and what is a B and what is a C and what is a D with the current reporting requirements. Basically, it says document your doses and report your doses. It still doesn’t cross into that new mindset from just having a range that you need to hit versus a target that you consistently hit. And so that’s what I would say as administrators and program leaders are working on this and there’s regulatory requirements are coming. It’s fine and it’s a good thing that there’s increased attention on this. So I wanted to say I’m really glad about that. But those regulations are not necessarily providing local sites with the guidance that they need to truly optimize their dose. In fact, it’s not. It’s based on the framework that people move from nothing to having kind of a range, a goalpost approach. Well, that’s better, it’ll improve, but it requires that mind shift to go from that goalpost approach to that target approach. And I would say, use this as an opportunity as you’re having to now get a handle on this anyway in terms of reporting your range as just take that extra step and now switch it to your target that you set the target and you hit the target. If you do that, then this actually could be a real opportunity to be transformational in the field. But I’m worried that that’s going to be hard for sites to do. They’re going to do the minimum, they’ll meet the requirements. And if they do that, I don’t think we’re going to have much of an impact.

Chris St. John – 00:37:01:

And so I just have to ask, what are some of the major misconceptions when it comes to Dose optimization, especially as reporting is going to be required soon? Like what are the things that people are getting wrong?

David Larson- 00:37:14:

Yeah. And I would say maybe I wouldn’t say getting wrong is, I think it’s a good question, but the reality is it’s not so much getting wrong. I think it’s more about what is kind of where the emphasis is and what people are focusing on that kind of tends to make it challenging. So I guess in terms of from a patient perspective or a public perspective, it’s kind of the assumption that doses are optimized and they’re not fully optimized. There’s room to improve for sure. But that increasingly is becoming, that awareness is continuing to be highlighted. I would say a couple of things that we tend to see in CT dose optimization by the professionals who do it. Number one is a focus on dose rather than on image quality. And of course, the whole point is to minimize your dose, unnecessary dose. So the thought is, well, if you can just measure your dose, your organ dose, and very sophisticated measurements of how much is each organ getting and how does that translate in an increased risk? And what is that cumulative dose over time? I mean, it’s well intended, but it doesn’t serve the purpose for optimization because optimization is really all about image quality, right? It’s really about getting that lowest dose for the needed image quality. That’s what as low as reasonably achievable means, really should continue that acronym should continue to say as low as reasonably achievable to achieve the target image quality. But that becomes a very… I love that.

Chris St. John – 00:38:35:

Like a supercalifragilistic style acronym.

David Larson- 00:38:38:

Yeah.

Chris St. John – 00:38:39:

We just got to keep extending it to get more and more specific and take all the nuance out of it.

David Larson- 00:38:46:

Yeah, that’s what it is meant by that. It’s like to do the job, right? As low as you can reasonably get to do the job. And so the job in this case is adequate image quality. That’s one misconception, or at least, you know, I’d say red herring that, you know, kind of misleads people. And then we’ve already talked about the kind of the goalpost versus the target approach. And then, you know, the thought that this is just up to the radiologist and the technologist and that the administrators don’t have a role in this. And they do. And I think that’s when it will get solved is when we see the managers and the directors of radiology departments with a reasonably sophisticated understanding that CT dose optimization means consistently hitting a target and they provision their department and set expectations that will enable them and hold them accountable for consistently hitting that target. As we see those are the, I say, the shifts that need to be made that will transform the field.

Chris St. John – 00:39:40:

That’s great. That’s a soundbite right there. So this one’s slightly selfish in nature, but as I am podcasting away here, learning what I can, what advice do you have for me going forward when I’m talking to people about Dose, when I’m talking to people about risk or lack of risk, when I’m talking to people about image quality? Do you have anything I can put in my back pocket? Any gems?

David Larson- 00:40:03:

I guess I would say I think the field is still converging. So you’re going to hear a number of different strategies, different philosophies, quite frankly. And I think we need to continue to work through those different approaches and converge on what is the approach we should focus on. I think that’s actually been one of the things that has challenged this field is there are so many different voices with different approaches. And so I leave it to you and you’ll have to figure this out and parse all that out. It’s hard to hear from different experts who say, you know, have very different philosophies. But I will say that my philosophy coming from a many decades in focusing on continuous quality improvement. I’ve become very firmly convinced of the need for set the target, hit the target, a target based approach for optimization. And personally, I think that the other methods, it’s not necessarily to criticize them specifically, but I see them as steps to get to this approach. And it’s very doable. I think often it’s thought, well, we don’t have the resources, we don’t have the time to do that. It’s actually quite doable. And that’s where the field is going to go anyway, in my opinion. So I hope we go there sooner rather than later.

Chris St. John – 00:41:11:

Yeah, I hope maybe this podcast can be the battleground around which different radiology professionals converge to fight it out. So if any listeners are ready to fight with Dr. Larson about his approach, please give me an email if you would like to talk.

David Larson- 00:41:28:

Hopefully this is a lover’s quarrel, shall we say, right? Meaning we all want the same thing, right? So nobody is as passionate as we are. In fact, that’s why we’re so passionate, because we all want the same thing. We want to minimize Dose. So even though we have different strategies, don’t let that sound like it’s not antagonism. It’s about striving to get it right.

Chris St. John – 00:41:45:

When it’s about discourse, right? It’s like the strength of an organization, the strength of a professional field is only increased by difference of opinion, conversation, discourse, critical thinking, conversation.

David Larson- 00:42:00:

And then hopefully then the data will be the deciding factor, right? At the end of the day, that’s what we should hope. We definitely need the discourse and the agreements and disagreements. And nobody agrees at first. You shouldn’t. In fact, that would be unhealthy if everyone initially agreed everything. But I would say at the same time, we should be moving forward towards convergence and with the data being the thing that moves us forward.

Chris St. John – 00:42:22:

Wonderful. Dr. Larson served as the acting chair and executive vice chair of the Department of Radiology at Stanford University for over 11 years. Dr. Larson, thank you so much for coming on to Frame by Frame Rethink Imaging. I had a wonderful time with you today.

David Larson- 00:42:36:

Thanks so much, Chris. It’s been an absolute pleasure. Really appreciate it.

Chris St. John – 00:42:38:

Really appreciate it.

Intro/Outro – 00:42:42:

Frame by Frame: Rethink Imaging is brought to you by Imologix. Here, you’ll find engaging interviews with thought leaders, experts, and patients, sharing stories that showcase the transformative power of medical imaging. To discover how Imologix is rethinking imaging in healthcare, visit imologix.com. Be sure to subscribe to Frame by Frame: Rethink Imaging on Apple Podcasts, Spotify, or wherever you listen. And from all of us here at Imologix, thanks for tuning in.