Rethink Imaging

EP 47 • June 25, 2026

Photon Counting CT Could Revolutionize Diagnostic Radiology

Featured Guest

Dr. Aaron Sodickson, MD, PhD

Division Chief of Emergency Radiology, Mass General Brigham; Associate Professor of Radiology, Harvard Medical School • Mass General Brigham

Listen Now40 min

Also on:

Let's Talk

Beyond dose

Ask us what your imaging data can actually show you.

Start a Conversation

Share This Episode

Subscribe to Rethink Imaging

Photon counting CT swaps the scintillator detectors that have defined CT since its beginning for semiconductors that convert each X-ray directly into an electrical pulse. Dr. Aaron Sodickson, who runs emergency radiology across nine Mass General Brigham emergency departments, explains what that change buys in practice: radiation dose cut another 30 to 50 percent, slice thickness down to 0.2 millimeters, and the energy of every single photon recorded. He traces the technology from the particle detectors at CERN that inspired it to the septa-free pixel design that lets resolution climb without raising the dose to the patient.

The second half of the conversation is about adoption, because hardware alone changes nothing. Sodickson has watched hospitals install high-end scanners and run them like fifteen-year-old machines, so he lays out what actually works: automated post-processing that pushes color-coded iodine images into PACS before the technologist finishes the exam, clinical champions paired with early adopter techs and medical physicists, and protocols that erased his pulmonary embolism scan failure rate. He also previews the DICOM standards effort he is driving with AAPM so radiologists can quantify iodine and fat content directly in PACS, the fix he calls the biggest lever for closing the adoption gap.

Host

Chris St. John

Host, Rethink Imaging • Imalogix

Featured Guest

Dr. Aaron Sodickson, MD, PhD

Division Chief of Emergency Radiology, Mass General Brigham; Associate Professor of Radiology, Harvard Medical School • Mass General Brigham

Watch the Episode

Key Takeaways

Photon counting detectors read the energy of every photon and discard electronic noise below a set threshold, cutting radiation dose another 30 to 50 percent below already-optimized levels.
Ultra-high-resolution mode images at 0.2 mm slices, down from the 0.5 to 0.6 mm standard, sharp enough to show individual bone trabeculae and catch subtle fractures.
Dual-energy iodine mapping lets radiologists clear many indeterminate kidney lesions as benign at the first scan, canceling follow-up imaging, extra cost, and weeks of patient anxiety.
Brigham and Women’s eliminated its pulmonary embolism CT failure rate using low-energy post-processing, then carried the technique back to its dual-energy scanner fleet.
Most sites that own dual-energy-capable scanners never use the capability. Sodickson’s answer is fully automated PACS workflows plus a clinical champion, early adopter techs, and medical physicists on every implementation team.

Full Transcript

Rethink Imaging Podcast Transcript

Guest: Aaron Sodickson, DMP, DABR

Host: Chris St. John

[00:01:48] Chris St. John: Dr. Sodickson, welcome to the show. It’s so good to have you here.

[00:02:01] Aaron Sodickson: Thanks, Chris. Nice to be here with you.

[00:02:04] Chris St. John: Let’s just dive in a little bit. You have held these leadership roles across clinical, academic, and tech-focused areas, and you spend a lot of time approaching and adopting, scaling new imaging technologies. I’m curious how your previous work has shaped your approach to the implementation of these technologies.

[00:02:34] Aaron Sodickson: There are basically a couple of roles that are most relevant here, I think. I spent about eight years overseeing our CT operations at Brigham and Women’s and many of our affiliated hospitals, and also spent about 15 years overseeing our emergency radiology practice as division chief first at Brigham and Women’s Hospital, and more recently also bringing Mass General into the fold to cover all of the nine emergency rooms in our network and a lot of urgent care centers.

And a couple of the things that are relevant here is that in emergency radiology, we are always under intense time pressure. We’re expected to provide very high-quality care on extremely short timeframes compared to most of radiology. This has taught me really to have a laser focus on efficiency and streamlined operations. On the CT side, my experience there really taught me about the importance of building the right teams for complex new projects with new technology like this. And in the case of what we’re going to be talking about today, dual-energy and photon-counting programs, it only works if you get the right multidisciplinary team in place, including your clinical champions from radiology, your enthusiastic early adopter CT technologists who are willing to try new things with you, medical physicists, the CT manufacturers, researchers. You got to bring all these people together when you’re trying to implement new things in a clinical space like this.

[00:04:06] Chris St. John: Did you found the CACTI organization? What is your role there?

[00:04:12] Aaron Sodickson: Yes, CACTI is the Center for Advanced CT Translation and Innovation. And this is basically a research group that I started about three years ago now. It’s aligned with where my CT interests have been for many years, which is really bringing the latest and greatest new technology into clinical translation, kicking the tires, figuring out where does it add more value for us? What should we do with it? Because the manufacturers create all this great technology and then people have to figure out what are we going to do with it? That’s not their job in the tech creation space. That’s our job on the clinical innovation side.

[00:04:55] Chris St. John: And something you’ve talked about with that group is the last mile translation. Before we get too inside baseball with dual-energy and photon counting, do you mind just explaining that? I feel like it’s an important piece of the puzzle before we get into the actual science.

[00:05:16] Aaron Sodickson: That term is relevant to a lot of new technology, not just in radiology, but anywhere. You’ve got really smart engineers, physicists, developers who make something and have ideas about how it might be helpful that they might test out in their lab or development environment. And then often there’s a gap when you actually introduce this into the real world, and in my space, into the clinical space for what can you actually do with it that you couldn’t do before? For that, you need clinicians—hopefully clinicians who also have an interest in the technology and understand the technology to figure out what to do with it.

After the manufacturers create all their great stuff, we like to bring it in and really make it sing in the clinical environment. How can we make it efficient? How can we add clinical value? How can we do better diagnosis or provide for better outcomes for our patients? All that stuff needs to be done after installing this new technology. It also creates challenges because it can be very difficult to put new and sometimes somewhat untested technology into a high-demand clinical environment because you still need to make sure that you can get all the work done. This is the space that I’ve occupied a lot with my research focus is figuring out how to best take advantage of this new technology.

[00:06:50] Chris St. John: It makes me think about when I’m working with generative AI, getting the first 90% of the way there can take minutes. It’s like, that’s so close to what I need and what I’m looking for. And then that last 10% ends up taking 99% of the time that you end up spending working on something. I think about that parallel. We have all of these different new technologies. They are some level of accessible, but how do you actually get them into practice?

[00:07:26] Aaron Sodickson: That’s a great analogy. You’ve got to learn how to prompt the thing the right way, how to actually get it to do what you want it to do. And the same is true for a lot of our advanced imaging technology.

[00:07:37] Chris St. John: Let’s jump into the advanced imaging technology. Dual-energy, let start there. I did go to your talk at RSNA last year. It was great. It was almost close enough to a year ago. Can you give me a little bit of a refresher course on the basics?

[00:07:58] Aaron Sodickson: Since I’m not sure exactly who the audience is for this, let me go even more basic and just start with what is conventional CT. In conventional CT, we’re basically passing X-rays through the patient and the equipment does a great job of calculating how well the contents that the X-rays pass through block X-rays. White stuff like the bones block a lot of X-rays, and dark stuff like air or fat or a lot of soft tissues block fewer of those X-rays. Conventional CT is basically looking at the ability of the materials in the patient to block X-rays.

The challenge there is different materials can block X-rays the same amount. If you take a head CT, for example, and you’ve got a bright spot in the brain, there’s a lot of overlap between benign calcification that’s been there forever and potentially a new hemorrhage. And we often on conventional CT cannot tell those two things apart and end up having to follow up the patient with another scan in six hours, something along these lines. It would be really nice to be able to differentiate not just on ability to block X-rays, but also on actual material information.

That’s what dual-energy starts to help us do. In conventional CT, we’re acquiring with a single energy spectrum of X-rays. In dual-energy CT, we’re acquiring with different energy spectra. All the manufacturers have slightly different ways of doing this, but at the core in dual-energy, we’re acquiring with high and low energy information. And different materials have different properties, different abilities to absorb high versus low energy X-rays differently. That means that if we acquire and post-process our images properly, we can tell that calcification from the hemorrhage. We can be definitive about our diagnosis right up front without having to do more evaluation.

[00:10:04] Chris St. John: What is the accessibility level of this right now? How abundant are these scanners, and broad strokes, are they being implemented nationwide? Is it a small little pool? Is it growing? Where are we?

[00:10:20] Aaron Sodickson: A lot of the higher-end CT scanners these days have the capability to do dual-energy acquisition and post-processing. The unfortunate reality is that most places that install these scanners don’t actually use it for these purposes. We have an implementation and adoption challenge beyond just siting the scanner in the hospital, and that gets back to what we started talking about a little bit about the teams that are needed here. You really need people to drive this forward because people are used to what they’re used to. Often what you see is people installing the high-end fancy new scanner and using it like the low-end basic scanner they had 15 years ago because there’s a learning curve, there’s decisions to be made about what we’re producing, and there is workflow optimization to sort out with the new post-processing. If you don’t do that work, then you’re left just using this more expensive higher-end scanner the same way you were using the earlier versions.

[00:11:33] Chris St. John: Got it. So it’s like the high school kid who gets into photography, gets a super fancy camera and just leaves it on automatic. I got to use analogies because it’s the only way I’m going to totally be able to understand what’s going on. When did you start engaging with this technology?

[00:11:57] Aaron Sodickson: We put in our first dual-energy scanner in the emergency room where I work at Brigham and Women’s Hospital in 2013. That was my start and did a lot of work up front there really getting into where we thought this could provide clinical benefit, diagnostic benefit for us, and sort of proved out a bunch of different areas all throughout the body that have become our standard for how we use this. One of the challenges with this is we’re all very busy, we’re all under intense time pressures. Anything you add has to be added for a reason, because otherwise you just further overwhelm our already overwhelmed radiologists who are trying to get through these scans. We were very careful to only add things that we had demonstrated were helpful. We can talk about various clinical examples at any point you want. But anyways, that’s when we started back in 2013. Spent about 10 years sort of building that dual-energy practice, getting everybody excited about it until photon counting came along as the next iteration.

[00:13:20] Chris St. John: Before we get there, I would love to hear some clinical examples. We can do that now.

[00:13:28] Aaron Sodickson: There are a number of different areas where we found this could be helpful. Broad categories: better ability to detect or characterize pathology of different types. Another category is making our imaging protocols more robust against failure. Another category: reduce radiation exposure by eliminating the need for multiple different exposures, multiple different acquisitions. And another category is reducing the need for follow-up imaging or other evaluation by more definitively characterizing pathology upfront when we find it.

There are a number of different materials that are great targets for dual-energy post-processing. I mentioned already calcium. Another one that we use all the time is iodine. This is the iodine that we give as part of the X-ray dye in our intravenous injections for CT. Those are a couple examples, there are others, but those are a couple of the most common examples that we can really do a great job of with dual-energy CT. In that head example I talked about, we can differentiate calcium versus hemorrhage in a trauma patient or a headache patient and say, “You know what, that’s just calcification, we’re done, go home.” As opposed to, “We’ve got to hold you here to make sure this isn’t a hemorrhage that’s going to expand.” That’s one use case.

Another one in the iodine space is the ability to see where the iodine is distributing, which tissues are taking up the iodine or enhancing as we call it. One example there is in the bowel. If we see a small bowel obstruction, one of the first questions is, is the bowel still alive? Is it still getting blood flow or is it dying or dead and we need to take it out? Very early on in our experience, the surgeons would start coming into the reading room with us and say, “I know there’s a small bowel obstruction, but show me that funny orange stuff you guys do.” The orange being the iodine that’s distributed throughout the body, which shows us that’s great enhancement of the bowel loop. It is alive, versus, “I’m really concerned about this bowel loop because the blood flow has been cut off as part of this bowel obstruction.”

The other area that I’ve been particularly excited about is incidental findings in the emergency department. We see a lot of new things that have not previously been imaged in patients. Kidney lesions is one of the common ones. So many patients have cysts. Many of these cysts are easy to tell because they have typical attenuation of water, but there are others that are complicated. They have a little protein in it, they have a little blood in it, and we can’t tell whether that’s a benign but complicated cyst compared to a tumor, cancer. So many of those patients were stuck. We have to recommend follow-up imaging even though we know the vast majority are going to be benign. The patient is stressing out that they might have cancer until they get that additional workup. It adds expense, it adds delay.

What we’ve shown is that for many of these lesions, if we can confidently say there’s no iodine content, no enhancement, no blood flow in that lesion, we can exonerate it as benign and never recommend a follow-up imaging, never stress out the patient, add that delay or expense. And there are other organs as well where we see a lot of these incidental findings that we can characterize much better with dual-energy CT. That’s one of the things that brings me a lot of joy when I’m reading these scans and I see one of these things that I know I would have had to recommend follow-up for, and I just never worry anybody, patient or ED doc for a moment that there’s something that we haven’t figured out because we can now figure it out.

[00:17:24] Chris St. John: Can I ask about the etymology of the term enhancement as it applies here?

[00:17:36] Aaron Sodickson: Maybe one way to think about it is if we had a CT scan without the CT dye, without the iodine administration, it would have a certain appearance. And when we give contrast and it gets brighter, it enhances the brightness or the appearance of the lesion based on the iodine that’s been taken up because of the blood flow to that area.

[00:18:00] Chris St. John: The word choice of enhancement is coming from the visualization when you’re actually reading the scans.

[00:18:08] Aaron Sodickson: Yeah. And as one example there, typically the way you would do the follow-up for those indeterminate kidney lesions is you will do a non-contrast scan, and then you would give the contrast, and you would do an enhanced scan, and you would look for differences in brightness or attenuation of that lesion after you’ve given the contrast.

[00:18:32] Chris St. John: I feel like I have a nice baseline understanding of what we’re talking about here. Can we get into how your workflow has changed as you started to implement?

[00:18:47] Aaron Sodickson: This is one of the big challenges when it gets to adoption because when we started, the only way to access the dual-energy information was we’d post-process the scan, we’d send the usual grayscale images to PACS, our image viewing system. But to look at the dual-energy information, we would have to go into the vendor-specific workstation or software to look at this additional information. In the early adopter days, I got all in on how to use that software and I convinced a handful of others on my team to do it, and most of the rest wanted nothing to do with that because we don’t have time to spend an extra five minutes on every CT scan just to unlock this additional information.

But over time and with a lot of nudging of the CT manufacturers, we were able to get to a fully automated workflow. What we can do is we can send the image data to the vendor’s post-processing software, have them automatically post-process it with no user having to touch it, and send it back to PACS with the rest of our images. Now what that allowed us to do is for every scan, in addition to the usual images that we interpret, we also automatically get these colored images that contain the iodine content flowing into our PACS system, usually before the tech’s even done finishing their work on the scanner. So we’ve got the whole package there.

The trick there is then deciding what extra stuff do we want to automate to send into PACS. As you’ve heard, we prioritized a lot of the iodine information. So we have these color-coded iodine overlay images where we have iodine in orange superimposed on our usual grayscale image, which makes it very easy to quickly see there’s iodine enhancement there and there and there, but not here. So we send those in automatically. And depending on the body region, we might have different post-processing that we’ve prioritized. Like in the head, we created a calcium versus hemorrhage application. So we color code the calcium to look different from the hemorrhage, and that allows us to do that differentiation.

[00:21:11] Chris St. John: In terms of the radiologists actually reading the images, using the language enhancements, talking about color, it feels like these images, in addition to just having more information that is legible, are they also just easier to read, or is learning to read them something else that has to be learned, studied, and takes additional time, skill, whatever?

[00:21:42] Aaron Sodickson: Yes, it requires a learning curve as well for how to use this extra information. The usual way this goes is people say, “Geez, that’s a lot. I’m already overwhelmed with my work. I don’t want more to look at.” And then what happens is you show them a couple of cases where they didn’t know, they couldn’t figure out which of these two options something was, and you say, “Well, you know what? We have more information. Why don’t you look over here at this image that we have provided for you for this purpose?” And you see the light bulb go on because people say, “Now I know what that is. I’m much more confident in my diagnosis of that thing.”

These other images are not a replacement for the standard grayscale images. They’re extra, and really, they’re there for problem solving and troubleshooting. So I don’t actually even expect people to look at them in every case unless there’s something they’re trying to problem solve where that additional information would be helpful. But it’s been really fun to see the adoption happen in that way kind of organically where people might not yet have much experience, but then they see a number of cases and they really start to get excited about this extra access of information that’s available now.

[00:23:07] Chris St. John: Y’all got your first dual-energy CT in 2013, 13 years ago. One of the other things I want to talk to you today about is photon-counting CT. Is it an evolution? Is it running in parallel? Or is it just an entirely different kind of subcategory?

[00:23:33] Aaron Sodickson: Photon counting, the change there is really that the detector, the CT detector has changed. You can still do dual-energy with that information; it’s just acquired differently. Should I take a moment maybe and talk about the detector differences?

[00:23:53] Chris St. John: Yeah, this is my sneaky little transition from focusing on dual-energy to taking up photon counting.

[00:24:00] Aaron Sodickson: Let’s do it. The CT detectors that we’ve been using since the beginning of CT have all looked pretty much the same. The X-ray comes through the patient and it hits this detector, and the first thing it hits is a scintillator material. The X-ray creates light in this material, and then the light hits an array of photodiodes that detect the light and turn it into signal. So the more X-rays that hit, or the higher the energy the X-rays that hit, the more light is produced, the bigger the signal is in that pixel. Because we’re using light, we have to have opaque septa between every pixel, which is fine unless you’re trying to get to really high resolution, because as you go to smaller and smaller pixels, these septa become proportionately pretty big and you lose a lot of radiation dose efficiency. You basically have to blast the patient with more radiation if you want to get adequate signal for the high-resolution task you’re trying to do.

[00:25:02] Chris St. John: Sorry, just to interrupt you, I am unfamiliar with the term septa, except for the public transit.

[00:25:09] Aaron Sodickson: What we mean by septa is we’ve got basically a little rectangle of this scintillator material, and we need to make a gap between it and the next pixel. A gap is manufactured, and it is filled with a material that blocks light. So that’s the septa or the septation, the septum that we’re talking about simply to prevent light from leaking from one pixel into the next, because if it leaks over to the next one, we’ve just lost spatial resolution. We’ve blurred out the localization of that information.

In moving away from those detectors, many years ago at CERN in Geneva, the particle physics place, they have these amazing detectors that detect the energy of the X-rays or gamma rays that come in and hit that detector. And people started to have the idea, “Well, this is the same energy range we use in CT. Maybe we can translate this type of detector to medical imaging.” And that kicked off a great collaborative effort between particle physicists and medical imaging device manufacturers and chip designers and physicists, engineers all over the place. And now, 20 years later, we have our clinical photon-counting CT scanners.

The difference now is that instead of that scintillator material to convert X-rays into light and then to detect how much light there is, these are made out of semiconductors that directly convert the X-ray that hits into an electrical pulse. So we skip the light and we have these little detectors that detect how much electricity, how many electrons are produced by each X-ray. A couple of nice features of this approach: first, we’re not using light anymore, so we can get rid of those septa between the pixels. We can eliminate the gap and we can make the pixels as small as we want, which means we can get to higher resolution than we used to be able to without increasing radiation dose. That’s one advantage. The other is that the size of the electron cloud that gets produced in the semiconductor is proportional to the energy of the X-ray. And so now we can directly detect the energy of every single X-ray that comes in. Can’t do that with the old, what are so-called energy-integrating scintillator-based detectors, where all the X-rays get sort of lumped together and we’re looking in aggregate at what the energy distribution is of the X-rays. So now we can detect each X-ray’s energy, which theoretically allows us to do these sorts of dual-energy applications a little bit more precisely.

[00:28:12] Chris St. John: At RSNA, you gave a bunch of examples of the vast array of benefits that photon counting yields. Can you talk us through, besides just reducing dose and increasing resolution, what are some of these other tangible benefits?

[00:28:30] Aaron Sodickson: Reducing dose and increasing resolution are good benefits. We’ve already been able to do really respectably low radiation dose imaging, but this allows us to knock it down even further—another 30%, maybe 50%, depending how aggressive you want to get.

[00:28:51] Chris St. John: Whoa, those are bigger numbers than I expected.

[00:28:54] Aaron Sodickson: Those are substantial numbers. And one of the reasons you can do this is that because they can detect the energy of each X-ray, you can actually make a threshold. And you can say, “You know what, everything with less energy than this, we’re just going to throw it out because it can’t have been a real X-ray that made it through the patient. That’s noise.” So we get rid of all of the electronic noise in the detector system, which is one of the reasons we can go to lower doses like this.

Anyway, radiation dose, yes, it’s a great benefit. High resolution does have applications where it’s fantastic. Musculoskeletal imaging is gorgeous. I absolutely love looking at these because now, typically we might be at a 0.6 or a 0.5 millimeter slice thickness. Now we have our standard mode is 0.4 millimeters. Our ultra-high-resolution mode is 0.2 millimeters. And so for all of our peripheral extremities, we’re looking at things with a 0.2 millimeter resolution. I mean, think about how small a millimeter is. We’re able to see such fine bony detail of all the little trabeculae running through the bones. Makes it easy to see even more subtle fractures that would otherwise be harder to find on thicker slices. So a lot of great benefits of the high-resolution imaging. And there are other places where people are excited about high resolution as well beyond just the bones: the lungs, the vessels in the brain, for example.

One of the areas that I really like about this is a lot of this relates to the dual-energy capabilities of these scanners. And so the same things we were talking about before, incidental finding characterization, so far it seems like we can do that even better with the photon-counting version of dual-energy CT than with the previous high-end version that we had before. Really just because the difference in the imaging chain allows for some more precise and accurate image post-processing at better resolution. So things like the kidney lesion characterization task, so far we’ve found that we can exonerate more of those lesions as non-enhancing benign things than we could do before.

Another area that’s been really nice here that we’ve used a lot is the ability to make more robust imaging protocols. One of the areas I’ve spent way too much time in my career doing is trying to optimize image quality of our pulmonary embolus CT protocols, and this is the protocol that we use to rule out blood or find blood clots in the lungs. To do that, you have to get a really well-timed scan to make sure that that iodine we administer is nicely distributed throughout the vessels in the lungs. And the timing can be pretty tricky. This is one of these exams that actually still these days in most places has a pretty significant failure rate where we just didn’t opacify the vessels as much as we needed to. We’ve used this technology to basically eliminate our failure rate on these scans. We use low-energy post-processing to really bring up the brightness of the iodine content that we got in and it’s made them incredibly robust against some of these timing challenges that I talked about. That’s something that we really spearheaded with this scanner, but we’re actually translating this back to our other dual-energy scanners once we realized what we could accomplish with these low-energy, dual-energy images.

[00:32:41] Chris St. John: What about other higher-level benefits, just around scanner utilization, repeat scans? I feel like there has to be some level of workflow efficiency that’s also coming out of these more controlled, more detailed images.

[00:33:01] Aaron Sodickson: The workflow is an interesting topic. It’s certainly got to be streamlined and efficient, and we’re using all the same tricks that we’ve used on our dual-energy scanners to fully automate as much of this stuff as we can do. At the same time, the scanner manufacturers have done a great job of developing other tools, some of them AI-based, some not, where we can more automatically align the scanning planes to the anatomy of the patient if they’re a little off-axis in the scanner. The other challenge is there’s a lot more data involved here because our pixel sizes are smaller. Inherently, the risk would be that this would take a lot longer, so the manufacturers have thrown much more computing power into their devices to overcome that.

And at the same point, if we want to use these higher-resolution, more-pixel images, we have to be a little bit thoughtful about how much we’re sending into our PACS system so we don’t crash our PACS system, for example, with images that are much larger per image times many more images because of the higher resolution. So a lot of considerations here. I’d say that the workflow isn’t inherently more efficient, but we’ve managed to keep it so that we haven’t had a loss in efficiency despite the higher-resolution, larger image stacks.

[00:34:35] Chris St. John: I want to circle back around to the work that you do with CACTI and this last mile as it relates to these technologies. Throughout the US, where are we kind of on implementation and utilization of these devices?

[00:35:01] Aaron Sodickson: This is a tough one, and I’ve been needling the CT manufacturers for years on this because transitioning from a dual-energy to a photon-counting scanner still doesn’t address our adoption problem with dual-energy data. All the things we were talking about earlier—the learning curve for the technologists, for the acquisition, the post-processing, how do we interpret the data, what do we send to PACS extra than we did before—all of those same challenges still exist. I think the manufacturers are getting better at creating more automated workflows sort of by default, but you still need the same implementation teams to get this into clinical practice. You need, most importantly, your clinical champions to really drive it forward with tight partnership with the technologists and the manufacturers and potentially the medical physicists at each place.

One of the things that I’ve been lately pushing on a lot with the manufacturers and the PACS vendors is we’d really like to see the ability to interrogate this dual-energy information in a quantitative way in PACS. Right now, what we send to PACS is basically a screen capture; I can’t measure on it. So I have a beautiful image that shows me the distribution of iodine content throughout my image, but I can’t quantify it. And some of the research tools that we’re creating, for example, for these kidney lesions or for adrenal masses that we find and we want to characterize, we need to be able to measure how much iodine is in it, how much fat is in it. And to do that, we’re stuck going back to the manufacturer software, which I’ve already told you nobody uses because we don’t have the time. We live and work in PACS.

Some of the CT manufacturers and PACS vendors have partnered to try to do this kind of thing, but it’s a one-by-one kind of partnership and it’s incredibly labor-intensive. So I’ve been trying to spearhead an effort recently to collect a lot of these people together and use some new DICOM standards that are under development to basically allow us to have the PACS manufacturers give us the display and measurement tools that we need to interrogate this in real time. And then we wouldn’t need to send all these extra images that we may or may not want in a particular case; we could do it all on demand in real time.

[00:37:54] Chris St. John: Are those DICOM standards something you and your teams are putting together or the manufacturers?

[00:37:49] Aaron Sodickson: No, it’s actually a separate effort. It’s got a combination of people from AAPM, the American Association of Physicists in Medicine. They’re great, they’re doing a lot of this work. They absolutely have people from the CT manufacturers’ teams on these DICOM standard committees, a lot of other medical physicists around in the community. And I’ve talked a lot with some of them and they’ve really developed some nice functionality specifically intended for dual-energy information to be able to access this in real time from the PACS setting. So I’m optimistic we’ll be able to wrangle everybody to sort of move forward in this way, because I think this would really be the biggest thing to help with the gap in adoption.

[00:38:44] Chris St. John: Right. And do you and your groups have resources available for folks who… it’s not like people are out there just buying photon-counting scanners like they’re nothing. Obviously access to the scanners and the machines themselves is a whole other can of worms. But for those who may have access and are not using them to their full capabilities, if they were to try and seek out educational materials, advice, guidance, where should they go to find that?

[00:39:15] Aaron Sodickson: At this point, there’s one manufacturer that has these out in the clinical domain, and they have been building a lot of educational content. We’ve also been sharing some of our protocols, and other people have been sharing some of their protocols that we’ve spent time developing and optimizing, so that people can have a pretty good starting point depending on what they want to do. So I think there’s a community certainly of expert users to help guide those that are newer in the space.

CHRIS ST. JOHN 00:39:50Last short, hopefully shorter little question before we wrap up. What about patient experience in these machines? Is it relatively one to one for them?

[00:40:00] Aaron Sodickson: I think most patients are probably unaware of what type of device we have them on, and maybe we could be doing a better job marketing this to our patients. But I think from a patient experience point of view, again, the part—I kind of mentioned this before—the part that makes me most gratified is knowing that I’ve been able to make a more definitive diagnosis without stressing a patient out with uncertainty. I love eliminating uncertainty with this technology. But most patients have no idea what specifics of the scanner we were putting them on.

[00:40:39] Chris St. John: And what about future evolution of CT and these technologies? Where do you see the direction of these advancements going?

[00:40:50] Aaron Sodickson: There seems, it seems quite likely that most of the CT manufacturers are really pushing heavily into the photon-counting space. I hope that that will mean that the cost of these scanners can come down right now. It’s really driven by the cost of the semiconductor for these detectors. And so the hope would be that at greater scale, these can become more accessible because these are really expensive scanners right now at this stage. If we’re going to have access more broadly to this technology, they’re going to need to figure out how to get the cost in line with most conventional CT scanners.

But I think there are enough benefits that make it reasonable for this to be the future direction of CT. We’ve talked a lot about dual-energy. I think there’s some AI hooks here as well—higher resolution, less noisy images with inherently more material content embedded in the information that we acquire. All great fodder for AI algorithm development, again, to do a better job of finding and characterizing pathology.

[00:42:13] Chris St. John: I would like to give a little shout-out to both of us, by the way, to do a radiology podcast specifically about emerging technologies and have made it almost the entire way through without one mention of AI. That is impressive and I appreciate it. We are always talking about AI on this show, our listeners know, but once again, I’m always clapping when we can have these conversations just like there’s just so much going on. And I love being able to dig into all of this technological information without just immediately always bringing the conversation back to AI, because we know that has its own can of worms when it comes to adoption. But anyway, not to dismiss you bringing it up, but I’m just saying proud of us.

[00:43:02] Aaron Sodickson: We did great. There’s one other, maybe one other thing I’d like to say, depending on who’s listening to the podcast. If we have any younger radiologists than this gray-haired guy giving the mic here, what I always like to encourage our radiology trainees to do is spend time learning the technology. So much of the focus in our radiology training programs is on interpreting images. There’s a whole piece that comes before that, which is acquiring the images and optimizing the image quality. And one of the best things I did years ago was sit with the CT techs and make them teach me how to run the scanner.

And the same can be said for MRI—learning the trade-offs. When you change this knob here or this setting, what happens to the images and what are the consequences of that? I think learning the technology and really understanding under the hood how it works makes us better at understanding what we’re looking at and makes it easier for us to do a better job. But I guess I’d also say that for anyone interested in the new technology or innovation space, to me this has been one of the most rewarding parts of my career because it’s new. You get to push the field forward by adopting new technology, by helping the manufacturers improve the technology to do what we want it to do. It ends up being a great collaboration, very rewarding and very fulfilling also to see other people that you’ve trained or that you’ve helped learn the tricks get excited about using this technology in our day-to-day. And it’s a really gratifying and exciting place to be, I think.

[00:44:54] Chris St. John: That’s a beautiful place to leave things. And I agree with you so much. I have a restaurant background, so I got to wear all of the hats. I was a chef, I was a server, I was a bartender, I was a dishwasher. To be able to understand the intricacies of all of these different roles and all of these different challenges is the most efficient and it’s the best way to be able to learn to speak the same language. Even within the world of imaging—and it’s one of the goals of this show—we’re sectioned off. We have physicists, we have techs, we have radiologists. And while we’re all working in the same space, I think so much of the nuance of our own different languages can be lost sometimes. And so I really appreciate that sentiment.

CHRIS ST. JOHN 00:5:50 Well, in that case, that is all the time we have for today. Dr. Aaron Sodickson, division chief of emergency radiology at Mass General Brigham and associate professor at Harvard Medical School, thank you so much for being here today. I really appreciate it.

[00:46:00] Aaron Sodickson: It was great to talk to you, Chris. I really enjoyed the conversation.

[00:46:07] Chris St. John: Thank you so much. My pleasure. Rethink Imaging is brought to you by Imalogix. 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 Imalogix is rethinking imaging in healthcare, visit

Questions from This Episode

How is photon counting CT different from conventional CT?
Conventional detectors convert X-rays to light and then light to signal, which requires opaque septa between pixels and lumps all photon energies together. Photon counting detectors are semiconductors that turn each X-ray directly into an electrical pulse, so pixels can shrink without a dose penalty and the scanner records every photon’s energy.
How much can photon counting CT reduce radiation dose?
Sodickson puts the reduction at another 30 to 50 percent below already-low modern doses, depending on how aggressive a site wants to get. Because the detector knows each photon’s energy, it throws out low-energy electronic noise that could not have passed through the patient, which drives much of the savings.
Why do so many hospitals underuse dual-energy and photon counting scanners?
Learning curves for technologists, unfamiliar post-processing, and time pressure on radiologists mean many sites run advanced scanners like the basic models they replaced. Sodickson’s fix is a multidisciplinary team with clinical champions, plus automated workflows that land processed images in PACS with no extra clicks.