Rethink Imaging

EP 29 • October 9, 2025

Detecting the Undetectable: Dr. Andrew Callen on Spinal CSF Leaks

Featured Guest

Dr. Andrew Callen, MD

Associate Professor of Radiology and Neurology; Director, CU CSF Leak Program • University of Colorado School of Medicine

Listen Now39 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

Spinal cerebrospinal fluid leaks are one of the most missed diagnoses in medicine, and Dr. Andrew Callen has built his career around finding them. In this episode he walks host Chris St. John through what CSF actually does, acting as the brain’s cushion and part of its waste-clearance and metabolic machinery, and what happens when that fluid escapes. Patients get a headache that eases when they lie flat and worsens when they stand, often with ringing ears, dizziness, and nausea. The classic story is only part of the picture. Many people leak without any of the textbook signs, which is exactly why they bounce through the system for years, misdiagnosed with migraine, and sometimes with far more frightening conditions.

Callen explains why a “normal” brain MRI does not rule out a leak, since a radiologist wrote normal based on what they saw and knew, not a lab value. He breaks down the specialized spine MRI protocol his team uses, the CSF-venous fistula that stayed invisible until 2013, and dynamic CT myelography, the study that catches leaks a routine myelogram cannot. He describes inventing the Myelovator, a hand-cranked positioning device that replaced blowing patients up on an inflatable transport mat. The conversation closes on the systems problem: this work loses his department roughly 60% of the RVUs a benchmark neuroradiologist brings in, and why he keeps doing it anyway for patients everyone else gave up on.

Host

Chris St. John

Host, Rethink Imaging • Imalogix

Featured Guest

Dr. Andrew Callen, MD

Associate Professor of Radiology and Neurology; Director, CU CSF Leak Program • University of Colorado School of Medicine

Watch the Episode

Key Takeaways

A normal brain MRI does not rule out a spinal CSF leak. Callen treats the brain MRI as a probability function for whether a search will find a leak, not a yes-or-no test, and reminds listeners that “normal” only means a radiologist wrote normal based on what they saw.
The CSF-venous fistula, the most common leak type Callen now finds, was only discovered in 2013 and first published in 2014. It produces no fluid collection in the spine, which is why patients with it were dismissed for years even when their brain MRI looked like a leak.
A routine CT myelogram and standard MRI show that fluid is leaking but not where. Dynamic CT myelography, with the patient injected and imaged in real time on the table, localizes the hole and can catch a CSF-venous fistula a conventional study misses.
Callen invented the Myelovator, a fiberglass frame with a hand-operated winch and a central hinge, to tilt patients in a controlled way. It replaced the older method of folding an inflatable hover mat under a patient’s hips and inflating it to raise them.
Running a CSF leak program costs money. Callen’s department calculated that pulling him from the reading room to do this full time yields about 60% of a benchmark neuroradiologist’s RVUs, so keeping the program alive took years of building a multidisciplinary financial model with hospital leadership.

Full Transcript

Rethink Imaging Podcast Transcript

Guest: Andrew Callen

Host: Chris St. John

CHRIS ST. JOHN 00:00:54 Today on Rethink Imaging, I am joined by Dr. Andrew Callen, Associate Professor of Radiology and Neurology at the University of Colorado School of Medicine. Dr. Callen is the founder and director of the CU CSF Leak Program, the first multidisciplinary center in Colorado dedicated to diagnosing and treating spinal CSF leaks. He has published more than 60 manuscripts, invented a novel patented positioning device for dynamic CT myelography, and serves on the medical advisory boards for both the US and Canadian spinal CSF leak foundations. He also directs the annual Spinal CSF Leak Bridging the Gap Conference, which uniquely integrates patient stories with physician expertise to advance care. Welcome to Rethink Imaging, Dr. Callen.

ANDREW CALLEN 00:01:42 Thanks for having me. Happy to be here.

CHRIS ST. JOHN 00:01:44 It’s so good to have you here. I just gave a lot of information in your bio talking about CSF leaks. Before we dive too far in, to catch me and potentially our listeners up, can we start with the basics? What is cerebrospinal fluid, and what role does it play in the brain and spine before we talk about leakage?

ANDREW CALLEN 00:02:11 That’s a good place to start. You can think of CSF, cerebrospinal fluid, as the brain’s natural cushion and plumbing system. The brain and spinal cord are floating in this fluid, encased in a lining called the dura. More recently in medical science, we’ve come to realize that it’s more than just a cushioning system. It’s actually a very important part of the way that the brain and spinal cord function. It’s involved in several metabolic processes, waste clearance, and I think that’s a good way to summarize it.

CHRIS ST. JOHN 00:02:58 Moving right on to the leakage, what is happening that is causing these leakages before we touch on how to find them?

ANDREW CALLEN 00:03:11 I think a good way to start this discussion is with something that people are more familiar with: a spinal tap, where a doctor puts a needle into the spine to sample some of that fluid for somebody who they think might have meningitis, for example. Doctors have known about CSF leaks in that context for a long time. They recognize that when we took that fluid out of someone’s spine, they would sometimes get a headache. It was a very characteristic headache, a headache that felt better when they were lying flat and worse when they stood upright. It tended to be in the back of their head and went into their neck. They often also had dizziness, ringing in their ears, and nausea. People came to understand that if that fluid is coming out, then people can get a headache.

It really hasn’t been until the last decade or two that an exponential increase in understanding of the concept of a spontaneous CSF leak has come around—where no doctor has put a needle in your back and you’ve not had a surgery that could have given you a CSF leak. We’ve come to realize that there are a lot of patients out there who develop these leaks and headache syndromes, but a lot of times not with that classic description I just gave, that are findable and fixable in many cases.

CHRIS ST. JOHN 00:04:40 How do we typically find these leaks? What imaging modality is used, and what does normal look like as we’re getting started?

ANDREW CALLEN 00:04:87 This is one of the most important parts of this discussion: where do we start? As radiologists or people interested in medical imaging, we’re not used to thinking as much about the clinical part of things, but that’s really where it starts. Somebody has to have the suspicion. For so many patients, perhaps they had a lumbar puncture, or a woman during childbirth had an epidural and the anesthesiologist noticed that there was a CSF leak during that positioning of the needle. But when it’s a spontaneous leak, the patient has to find their way to a provider that recognizes, “Wait a second, this could be a secondary headache disorder, not a migraine,” and then hopefully orders neuroimaging.

In general, we want to look at a routine MRI of the brain—nothing super fancy. We believe that most patients with a CSF leak will show changes on that brain MRI reflective of a spinal fluid leak. Those changes fall into two pathophysiologic buckets, if you will: one is that the brain will sink or sag downward, and the other is that the venous structures or blood-filled structures will engorge to take up the space left by the CSF leaving the brain and spinal compartment.

The problem is that there is some portion of patients—that percentage is debated—who will have a normal MRI of the brain despite an active CSF leak. This makes it even more challenging because even if somebody thinks, “Could this be a leak? I should get a brain MRI,” if the brain MRI is normal, perhaps they won’t ever make it to the next step of a workup.

What’s even more interesting when you think about the word “normal” is: what is a normal MRI of the brain? It’s simply one where the radiologist wrote “normal” in the report. One of my favorite sayings from one of my old teachers when I was in training was, “Radiology is not a lab test.” It’s not a machine that’s counting something and putting it out; it’s a person looking at images and interpreting them based on what they know, what they’ve learned, their biases, et cetera. So you wonder, among this percentage of people that we say have a normal brain MRI, what percentage of those have subtle findings that were just not picked up on? Then that “normal” report is in that patient’s chart, part of their medical record, and stuck with them. If you’re not being cared for by somebody who recognizes that despite normal imaging, they could still have this condition, their workup often stops there. That’s the first line of how things tend to go.

CHRIS ST. JOHN 00:08:05 Let’s say you have a patient who comes in with a headache disorder. Maybe they’ve tried Botox injections or something to handle it, they’ve had an MRI, everything looks normal, and the headaches aren’t going away. What is the next step in trying to identify a leak?

ANDREW CALLEN 00:08:25 Our referrals come from three ways that patients make their way to me. One is they’ve been working with a neurologist or a headache doctor who has been trying to address primary headache disorders—migraine, tension headache, cluster headache. They’ve been trying medications and interventions directed at those disorders, like Botox, and it’s just not working. There are migraines that are refractory to treatment, but that is one way: the neurologist reaches a dead end and wonders, “Could this be a secondary headache disorder?” Or they come in with a classic story for a CSF leak, and the neurologist says, “This sounds like a CSF leak. Let me work you up and then refer you to somebody who could further investigate this.”

The next way is that the patient advocates for themselves and self-refers to our program or another program because they haven’t been able to get an answer for their symptoms, do some research online, and think, “This sounds like what I have.”

Then we have people who are getting worked up for something else, and incidentally the imaging shows signs that could reflect a CSF leak. If the radiologist is in tune with this diagnosis, they say this and the patient gets funneled in. For those patients, a lot of them have what we would call atypical presentations—maybe not a classic orthostatic headache. When their primary care or whoever ordered the study asks, “Do you have a headache that’s better when you lie down?” and they say no, the provider thinks, “Okay, this can’t be a CSF leak.” But that’s a big diagnostic problem because many of these patients don’t have that classic story.

Once they make it to us, we get an MRI of the entire spine. The vast majority of CSF leaks that result in this condition occur in the spine. CSF leaks can occur in the skull base too, but headaches that are worse in the upright position occur in the spine. So we examine the spine with a special MRI protocol. A routine MRI protocol is fine for identifying a big epidural fluid collection, but we like to get a more detailed MRI protocol consisting of 3D heavily T2-weighted—meaning really focusing on the fluid signal—fat-suppressed imaging that allows us to scrutinize the dura, the nerve root sleeves where many of these leaks occur, and subtle findings related to a prior iatrogenic (doctor-caused) injury, a tiny irregularity where a needle may have been, subtle fluid, and so on.

CHRIS ST. JOHN 00:11:19 When you were talking about the specialized protocol for the spine MRI, you mentioned fat suppression. What is that concept within the context of protocols?

ANDREW CALLEN 00:11:40 Starting from the basics, with a CT scanner or an X-ray, you put the patient in, press a button, and it takes a picture—it’s always the same kind of picture. For MRI, we have to ask specific questions. We have to say, “I want to look at the bone,” or “I want to look at the fluid,” and it relies on the characteristics of the tissue you’re looking at.

When you take a picture with T2-weighted imaging, that makes fluid look bright, like CSF, but fat also looks bright—just normal fat in our body. There is fat surrounding the dura in the epidural space. If you get imaging that does not make that fat go dark while keeping the fluid bright—that’s fat-suppressed imaging—then you could mistake fat for fluid and wonder if someone has fluid outside of the dura when in reality it’s fat. So we make that signal go away with that protocol.

ANDREW CALLEN 00:12:44 We have the MRI of their brain and the MRI of their spine, and then we meet them in clinic and talk to them. A lot of these patients have been through the medical system, navigating and seeing lots of specialists with different diagnoses considered. When I meet a patient, I want them to start from the beginning—pretend like I don’t know anything, have no biases, and just tell me the story from where it began. Sometimes you’ll hear the same thing that’s in the chart; sometimes you’ll discover something new. What we’re looking for is a sudden-onset, daily, new headache, usually associated with vestibulocochlear symptoms—meaning things related to balance, ringing in your ears, a feeling that your ears need to pop or that hearing is muffled, vision issues, et cetera.

We get the whole timeline down. We want to know if they’ve ever had a needle in their back, even if they don’t think it’s related—any spinal tap or epidural at any time—and really understand the timeline. It doesn’t necessarily need to be that their symptoms started right after that intervention.

Then we look at their imaging together. For patients who have subtle imaging or even negative imaging, I explain to them that 10 or 20 years ago, no one looking at their brain MRI would think there’s a chance they have a CSF leak. But instead of thinking of the brain MRI as being diagnostically binary, I think of it in terms of a probability function—a function of the likelihood that if I look for a CSF leak, I will find one. If the brain MRI shows lots of findings of a CSF leak, there’s a good chance I’m going to find one. If it doesn’t, then it is unlikely, but not impossible, that I will find one.

When we look at the spine MRI, if the patient has a big fluid collection, it’s a slam dunk: there’s a CSF leak, and we just have to figure out where it is. But if they don’t have a fluid collection, we wonder if this could be a specific type of CSF leak called a CSF venous fistula. A fistula is an abnormal connection between two things, and in this case, it is between a vein next to the spine—a normal vein plugged in and sucking fluid out into the blood—rather than a hole with fluid pouring out. This is an incredibly important concept because this type of leak was only discovered and described around 2013 or 2014, so it’s a medical diagnosis that’s only 11 years old.

When these patients have a leak, they do not have a fluid collection in their spine. That is one of the reasons this type of leak evaded detection up until that point, and people were dismissed: “There’s no fluid, so it can’t be a leak.” Further, if we just do a regular myelogram—where we inject X-ray dye into the spinal fluid, roll the patient around, and put them in a CT scanner on their back 30 minutes later, which is done at almost every hospital in the country—you will not see a CSF venous fistula. You have to do a special type of myelogram where the patient is on their side, injecting contrast dye that layers densely along the side of the spine. Sometimes we have to do provocative maneuvers, such as changing how they breathe or raising their intrathecal pressure to see these fistulas, so it’s a very involved study.

This is why we talk a lot about the probability of finding a leak with the patient. In the case where their brain MRI is subtle or normal, I tell them very frankly, “I have found CSF leaks and fistulas in patients with this type of brain MRI, but the odds are against us.” The catch-22 of my job is that in order to find a CSF leak with a myelogram, I have to put a needle through someone’s dura. The last thing I want to do is give the patient the disease I’m trying to treat. Not to mention this exam involves a good amount of radiation—we’ll usually do at least two passes through the spine on each side, so four whole-spine CTs, sometimes more. For younger patients in particular, the last thing I want to do is perform this test, give them a post-dural puncture headache, and increase their long-term cancer risk without helping them.

It’s very important to have this conversation about risks and benefits. If the brain MRI is super positive, it’s a no-brainer. But if it’s subtle, we have that conversation.

Depending on whether we find something during that myelogram, that’s great. It’s almost like magic to find these fistulas, particularly for patients who have been suffering for a long time. To show that to them is incredibly validating, and then we go about treating it.

In patients where we don’t find something, we discuss whether it’s reasonable to try an empiric therapy, meaning we offer the patient a blood patch even if we don’t see a leak. A blood patch involves taking the patient’s own blood from an IV and injecting it into the epidural space outside the dura in the spine. Sometimes this helps—sometimes temporarily, sometimes long-term or permanently, completely or partially. We try this because no other therapy is working. It is an invasive procedure, but it does not involve a dural puncture, uses much less radiation, and is low-risk.

These are the two ways we think about the initial branch point: should we do a myelogram, and if we find something, great; if we don’t, should we offer empiric treatment? Or, if someone is young with a brain MRI that isn’t showing obvious signs and no other treatments have worked, maybe we start with a blood patch to see if it helps. If it does, that gives us a clue and adjusts the risk-benefit ratio for a myelogram.

CHRIS ST. JOHN 00:19:20 Not to backtrack too much, but when you were talking about fistulas specifically and iatrogenic causes of CSF leaks, do those run in parallel, or could an iatrogenic cause lead to a fistula, or are fistulas universally spontaneous?

ANDREW CALLEN 00:19:57 The general basic heuristic is that spontaneous leaks are just that—spontaneous, not caused by an intervention. However, a paper published last year described patients undergoing an intervention that could result in a spinal CSF leak—a lumbar puncture, an epidural steroid injection, or spine surgery—and experiencing symptoms suggestive of a leak. When referred for a workup, a leak is found, but it’s not where the surgery or needle was; it’s several levels away, and it’s a spontaneous leak, not a hole caused by a needle.

How does that happen? Is there something triggered by an intervention in the spine even far away, or is it anchoring bias on the part of the clinician because the patient had a procedure? A patient might have had neck pain from an underlying CSF leak, received a steroid injection that didn’t help, and actually had a CSF venous fistula the whole time.

Furthermore, there’s crossover in that it’s been observed that following an injury to the dura—trauma, surgery, or an iatrogenic needle—a fistula with an abnormal vein can grow there and drain fluid. One of the most challenging patient groups is those with chronic post-dural puncture headache. We were taught in medical school that a post-dural puncture headache is a benign, self-limiting phenomenon—have them lie down, do a blood patch, and they’ll be fine. But these patients can suffer immensely for years after a lumbar puncture. When you look at their spine, there’s no fluid collection and no obvious findings on brain MRI. I’ve wondered if a fistula forms where that puncture was, or if there’s vascular changes we’re not yet able to see in the same way we couldn’t see fistulas until 2014. There are many questions that remain to be answered, and the crossover between iatrogenic and spontaneous causes is very interesting.

CHRIS ST. JOHN 00:22:25 Is there a delay phase sometimes between when they have a procedure and when symptoms of a leak start? Can it be months or years?

ANDREW CALLEN 00:22:40 Yes, absolutely, for several reasons. We’ve taken care of and published on patients where this occurred. One patient had a labor epidural and years later developed a leak, and we found an injury near where the epidural needle was. Why did that occur? Was it worsening of the leak over time, or a slow opening? Did something else happen physiologically that increased pressure on that area? In the case of surgery, if the patient had bleeding that acted like a blood patch, there might be a delay until that blood resorbs and the underlying leak reopens. Or is there a compensatory mechanism keeping people okay for a while until it fails? This is a huge diagnostic problem—if it’s not a sudden onset right after exposure to a risk, people have a hard time connecting those dots.

CHRIS ST. JOHN 00:25:01 To get back into the imaging process, dynamic CT myelography has been central to the work you’re doing. What is different about that from traditional myelography?

ANDREW CALLEN 00:25:30 A good place to start is talking about digital subtraction myelography (DSM), which was the technique that really propelled the field forward, pioneered largely by Dr. Wouter Schievink and Marcel Maya at Cedars-Sinai. They realized that if you use the technique an interventionalist uses during a catheter angiogram—taking live X-rays, subtracting the background out, and injecting dye—you can see just the dye over the subtracted image in real time. If there’s a hole in the dura, you will see the dye come out. That is a digital subtraction myelogram.

With a conventional CT myelogram, you inject dye under X-ray, take the needle out, have the patient roll around, and put them in a CT scanner 30 minutes later. If there’s a hole in the dura and fluid in the epidural space, dye fills that space and the CSF, so you see that the patient has a leak, but you have no idea where it’s coming from or where the hole is. It’s essentially equivalent to an MRI—you see fluid, but not the source. Furthermore, you won’t see a CSF venous fistula.

Shortly after DSM was described as a tool for localizing dural holes because of its temporal resolution, people realized we could do this on CT as well. CT offers major benefits: we can look at the entire spine in one run, whereas with fluoroscopic DSM, you look at one station at a time. With CT, we can also evaluate both sides of the spine in one day; with DSM, once contrast is in the spine, a second injection makes images very difficult to interpret.

So people thought, why not use these same techniques on CT? Instead of injecting under fluoroscopy, waiting 30 minutes, and scanning, what if we inject dye while the patient is on the CT table and take fast pictures as dye moves down the spine? We achieve what DSM does, but we can evaluate the whole spine at once, including areas that are tough on DSM, such as the high thoracic spine across the shoulders where superimposition artifacts occur. We can also see contributory anatomy, like a bone spur or spicule that poked a hole right next to the leak, and we can plan percutaneous image-guided therapies right on the table.

The downsides are temporal resolution—it’s not a continuous video like fluoroscopy—more radiation exposure, and the angling issue, since CT tables don’t tilt.

CHRIS ST. JOHN 00:29:45 How do you angle the patient on a CT table?

ANDREW CALLEN 00:29:47 I was taught to do this by Dr. Bill Dillon at UCSF, an early pioneer in this field and a close friend and mentor. He taught me to use a HoverMatt—a patient transport device, an inflatable mattress used to move patients. We would fold that in half under the patient’s hips, inflate it so their hips popped up in the air, let dye flow down, deflate it, and image. It worked well, but it’s intimidating for a patient who has waited months for an appointment to see you blow them up on a balloon mattress.

Most people use a foam wedge on the table and position the patient on their side or face down. That can work, but there are problems: it’s not a one-size-fits-all solution for tall patients, and if a patient has low pressure and you angle their hips up while attempting a lumbar puncture, you decrease tension on the thecal sac, making needle insertion challenging. You also cannot reliably measure opening pressure in that position.

Having a way to start flat, perform the lumbar puncture, and then raise the patient in a controlled, safe way is imperative. That’s why we developed a patient positioning device for the CT table called the Myelogram Elevator. It’s a fiberglass frame with reinforced polymers and a hand-operated winch. As I crank the winch, a strap pulls one side of the table close to the other around a hinge in the middle, allowing us to tilt the patient in a controlled, precise manner. I use this every day and have performed hundreds of exams with it. I can keep a needle in while flat, slowly raise the patient, and adjust the angle as needed without sacrificing exam quality.

CHRIS ST. JOHN 00:34:24 You had to invent your own device to image patients the way you need to, so running a CSF leak program is very different day-to-day compared to reading routine neuroradiology studies. How has this specialization reshaped your workflow as a radiologist?

ANDREW CALLEN 00:34:40 System-related issues are one of the most critical obstacles to advancing this field. Radiologists generally don’t go into diagnostic radiology to run clinics or do procedures all day. Furthermore, the economics are challenging: neuroradiology is a well-reimbursed field in terms of RVUs per study, supporting the department financially. Taking me out of the reading room to run clinic or perform time-intensive procedures—which lack advanced procedural billing codes—reduces RVU generation. When I perform a dynamic myelogram, it is billed like a regular myelogram. My department calculated that transitioning me to full-time CSF leak work results in generating about 60% of the RVUs of a benchmark neuroradiologist, despite working 10-hour days from 7:00 AM to 5:00 PM without breaks.

As a specialty, we need to establish appropriate billing codes for this work. To be successful, we’ve had to think outside of radiology and collaborate with neurosurgery and neurology. Patients with CSF leaks engage multiple dimensions of the health system. I engaged hospital leadership to build a multidisciplinary financial model that compensates for the new patient volume brought into the system while supporting the department.

It took years to establish this model, and it remains an ongoing effort that I discuss with colleagues at other institutions. In private practice focused strictly on RVU volume, taking someone off the reading schedule for specialized procedures is difficult. We must overcome these logistical obstacles for the field to advance. In five to ten years, I expect most tertiary academic medical centers will have dedicated CSF leak programs because these patients are out there.

CHRIS ST. JOHN 00:39:30 The first description of a CSF venous fistula was published in 2014. Where do you see the field and specialty growing over the next five to ten years?

ANDREW CALLEN 00:39:40 The first step is recognition. When the CSF venous fistula was first described, people thought it was a rare entity. Now, fistulas represent the majority of leaks I find. These patients are everywhere, often misdiagnosed with migraines or other pain syndromes. Recognizing that this is not unusual is number one.

Number two is keeping an open mind about what we don’t yet understand. The most rewarding part of my job is finding someone’s leak—especially when imaging was subtle—validating their experience, and helping them. Conversely, the most frustrating part is when I can’t help someone whose story sounds like a textbook leak, but imaging and empiric therapy reveal nothing.

There is a subset of patients for whom we do not yet have answers, who suffer in silence and engage in unproven therapies out of desperation. Because their clinical presentation resembles a leak so closely, I believe there is a major pathophysiologic mechanism we are missing regarding the cerebrospinal fluid and surrounding venous compartment—whether another leak subtype or a leak-adjacent condition. Finding an answer for those patients is our job moving forward.

CHRIS ST. JOHN 00:42:18 Thank you so much for joining us, Dr. Callen. Dr. Andrew Callen is Associate Professor of Radiology and Neurology at the University of Colorado School of Medicine and Director of the CU CSF Leak Program. Dr. Callen, thank you for joining us on Rethink Imaging.

ANDREW CALLEN 00:42:36 Thanks for having me, it was my pleasure.

Questions from This Episode

Why can a patient have a spinal CSF leak but a normal brain MRI?
A brain MRI usually shows two patterns when someone is leaking: the brain sags downward, and the venous structures engorge to fill the space the fluid left. Some patients show neither, and the report reads normal. Callen points out that normal only means the radiologist did not flag anything, so subtle findings can be missed and the workup often stops there.
What is a CSF-venous fistula and why was it missed for so long?
A CSF-venous fistula is an abnormal connection where a vein next to the spine draws spinal fluid directly into the bloodstream, so there is no fluid collection to see. It was only described around 2013 and 2014. Because a routine myelogram cannot show it, patients with a fistula were told they could not have a leak because there was no visible fluid.
How does dynamic CT myelography differ from a conventional myelogram?
In a conventional CT myelogram, the dye is injected, the patient rolls around, and the scan happens about 30 minutes later, showing fluid but not its source. In dynamic CT myelography, Callen injects the dye with the patient already on the scanner and takes fast images as it moves down the spine. That timing lets him localize the leak, image the whole spine at once, and catch fistulas.