The O.G. fatal MRI accident – that reshaped MRI safety best practices around the world – happened in 2001 at Westchester Medical Center in New York. This second series from Invisible Force takes you back to that accident, recreating the accident that wound up killing six-year-old Michael Colombini. While many in radiology know the headline of the most infamous deadly MRI accident, surprisingly few know what happened or what went wrong on this, the 25th anniversary of the incident.

This episode tells the story of how the young boy found himself inside the MRI scanner, and the sequence of events that turned what would have been a miraculous diagnosis and surgery into a tragedy that snuffed out Michael Colombini’s life.

After this episode, the next two episodes in the series will look at the near-term implications, including the New York State Department of Health (NYSDOH) investigation into WMC and the lawsuit, and then the longer-term fallout from this accident that shocked the conscience of the radiology profession, looking at the development of best practices and the sorry state of MRI safety licensure or regulation minimums. Please subscribe to make sure you don’t miss a single episode!

Show Notes: Westchester County Clerk’s Office (Case Index No. 11101/2002)

Transcript:
Help!

My patient is de-satting and the inner oxygen isn’t working. My patient needs oxygen.

What did I see? The portable oxygen cylinders?

Oh, yes.

Hold on, doc. Here’s an O2 tank.

Thank you.  [CRASH]

Hello, and welcome to the second series of the Invisible Force Podcast. This podcast series is built around exploring MRI accidents and incidents that often get described incorrectly in local news or in online stories as being freak accidents. This second season is dedicated to an MRI accident that really launched contemporary MRI safety 25 years ago in July of 2001. This accident occurred at Westchester Medical Center, and over the next three episodes, we’re going to describe for you the tragedy that took the life of six-year-old Michael Colombini and how this accident became something of a watershed moment for MRI safety. But before we launch you into the second series of the Invisible Force and the story of Michael Colombini, let’s reintroduce our hosts.

I’m John Posh, MRI technologist, educator and long-time advocate of MRI safety practices. I’m also adjunct faculty at two universities and teach MRI safety multiple times per year.

And I’m Toby Gilk, certified MR safety officer, certified MR safety expert, MRI facility architect, safety consultant and co-author of a new MRI safety textbook, the Technologists MRI Safety Handbook.

In our first series, we put on our goofy-looking Sherlock Holmes hats and were investigators into the 2025 death of Keith McAllister. This series looks at the OG MRI incident that most of us are aware of, the 2001 death of Michael Colombini. Now for the Colombini accident, much of the investigation was done for us and is all part of the public record from litigation that followed. Like the McAllister incident, this earlier incident occurred just outside New York City and got a ton of press coverage, probably even more than the recent death. This year, 2026 marks the 25th anniversary of this accident, and many in health care have never heard the full story, and so we’ve lost touch with any lessons that this accident can teach us. Our look back at this accident is both because this milestone anniversary seems like an appropriate time to reflect on it, and because it is such a seminal moment in MRI safety that being familiar with it, really helps the understanding of everything else that happens after.

It’s also worth noting that we treat Colombini as the first fatal accident. There were others before, at least one similar accident that occurred a few years prior in Texas. The key difference with the other serious MRI accidents that happened before Colombini is that they were effectively kept out of the press. As our series continues, this will turn out to be one of the most important internalized lessons about MRI accidents that health care providers take away. Silence, deny and obfuscate. For better or for worse, you’ll learn that that’s not what happened in this specific incident.

The most famous American MRI accident begins, ironically, with a CAT scan. But hold on, I’m getting ahead of myself. Let’s hop in the Wayback Machine and go back to the summer of 2001 in the bedroom community of Valhalla, New York. Valhalla is a suburb of New York City, and lots of people live here and commute into the city for work. One fateful summer day, a rambunctious six-year-old boy, Michael Colombini, had a ton of energy, so his mother took him to a neighborhood playground to be able to burn off some of that pent up energy.

I imagine that his shoelaces wound up coming untied while he was running around, and he winds up tripping and landing hard on the pavement. His mother rushes to Michael to help him up. He hit his head when he fell, and he’s now wobbly and dull. He says his head hurts, and very shortly later, he says that his stomach doesn’t feel right. Mom recognizes these as all possible signs of a concussion, and she wastes no time in taking him to a local community hospital ER. In some ways, what happens next is really the beginning of the story.

At that community hospital, they were worried about the possibility that young Michael had fractured his skull, so they did a CAT scan. That same CAT scan I mentioned just a moment ago. What it told doctors at the community hospital was the classic good news, bad news diagnosis. The good news was that the CT showed that Michael did not have a skull fracture. Yes, it seemed that he had a concussion, but there was no long-term damage from the fall. Now, the bad news, however, must have completely blindsided Mrs. Colombini. The CT showed that her son had a brain tumor.

Now, this particular type of brain tumor is called an astrocytoma. This tumor gets graded on a one to four scale, where one is very treatable, but anything two or higher have worse and worse possible outcomes. Grade four of these tumors in 2026, not 25 years ago, but now, grade four tumors have a mean survival time of one to one and a half years. For those patients whose tumors are found because they’re far enough advanced that they’re exhibiting symptoms, well, those patients don’t tend to do very well. Because he was so young and because the tumor was small, this discovery was really something of a miracle. This allowed an intervention to save this boy from a horrible discovery in his young adulthood. But this community hospital didn’t have neurosurgeons who could treat this, so they wound up transferring him immediately, do not pass go, do not collect $200, to the larger hospital, Westchester Medical Center, where the boy could have the indicated brain surgery.

So, the Colombini family takes Michael to Westchester Medical Center. Up until a few years prior, it had been a county-owned hospital, but it had been transferred to the Public Benefit Corporation in 1998. This is a big hospital with lots of types of specialty care available, including neurosurgeons specifically trained to operate on brain tumors. But it’s not as if they whisk him directly into the OR. This tumor isn’t so aggressive that a day or two is likely to make any difference, and this is the type of surgery that needs to be carefully planned. But at the same time, they wanted to act sooner rather than later. One of the ways they planned the operation was to collect more imaging to better identify where the tumor was, and just as importantly, where the surrounding healthy brain tissue was that they didn’t want to disturb or remove. So they sent Michael to a pre-op MRI.

This pre-op MRI is not the MRI that this story is about. That one, the pre-op was completed without incident. Between the CT study from the community hospital and the pre-op MRI completed at Westchester Medical Center, the neurosurgery team could figure out where the tumor was, how large it was, and the degree to which it might be invading nearby healthy brain tissues. With this, they could decide where would be best to cut open what’s called a bone window in the boy’s skull, a hole large enough to get surgical instruments into his brain in order to remove the tumor.

If you’re going to have a surgical procedure that requires anesthesia, one of the greatest risks to the patient is vomiting. To reduce this risk, normal anesthesia preparation protocols are nothing to eat or drink for at least eight hours prior to the surgery. I try and put myself in the position of Colombini’s. Michael is still in pain from the fall, told that he has a brain tumor that could potentially kill him if they don’t operate and get it all. The boy is getting his head shaved, all of them stuck in a hospital, ticking off the hours until brain surgery, without even the ability to comfort him with a snack or a drink. The stress and anxiety would be off the charts for me if I was in that situation.

On the second day in the hospital, what must have felt like a decade for the family, they get the boy into the operating room. The surgeons do, in fact, saw out a window from the boy’s skull, carefully setting the piece of bone aside to be able to put it back in with some screw anchors after the surgery is complete. They navigate through the healthy brain tissues, gently pulling apart the folds to uncover the location of the tumor. Referring to the imaging, because tumors don’t always appear different from the nearby tissues, the surgical team identifies what they want to cut out and what they want to leave undisturbed.

The imaging is good, and the surgeons have a good indication of exactly where the tumor starts and stops. They’re able to access the spot without too much difficulty, and by all accounts, the surgery goes very well. The surgeons are confident that they got the whole tumor. They replaced the rectangle of bone they cut out of his skull, and with a handful of little titanium tabs to hold it in place, put screws through the tabs into the rectangle of bone to secure it in place. A dozen surgical staples to reconnect the flap of scalp that had been pulled away, and it looks as if he’ll be good as new after he heals from the surgery.

The news from the neurosurgeon to the family must felt like the weight of the world was lifted from them. The surgery went very well. They believe they got the full extents of the tumor, though to be certain, they’re going to want to do an MRI on Michael regularly and look for any sign of recurrence. But the news that day was all good. The playground accident a couple of days prior turned out to have possibly saved their son from a horrible death three, maybe four decades later. Their son, while looking a little bit worse for wear after his surgery, Michael was going to be fine. In that moment, they must have thought that they’d just close this chapter and that the worst was behind them. In that moment, nobody was aware that their moment of relief was going to be very short lived and the next day was going to change their lives. And as it happened, the very trajectory of MRI safety. 

The sponsor of the Invisible Force podcast is cairereporting.org. That’s CAIRE, spelled C-A-I-R-E. carereporting.org is a confidential MRI adverse incident reporting system available to you, the public, as well as offering an enterprise-level solution to hospitals and imaging centers for secure, confidential reporting of MRI incidents, accidents and near misses. If you have direct knowledge of an MRI accident, that may have been swept under the rug. Or if you want information about how CAIRE could be set up as a private internal error reporting tool for your hospital or imaging center, CareReporting can help you with both of those. CAIRE has an assembled panel of international experts in MRI safety and accidents. Reports submitted either through the public website or through an enterprise level system get reviewed by their experts who then deliver insights into the contributing causes of how accidents or near miss events happened, and more importantly, steps that could prevent similar incidents in the future. For more information, please visit cairereporting.org. That’s CAIRE spelled C-A-I-R-E.

Just before the break, we left you with a deeply relieved Colombini family. They had just reached there, and they all lived ‘happily ever after moment,’ hearing that the operation to remove the tumor from Michael’s brain was a huge success. They’re still in the hospital with an exhausted Michael on painkillers, but able to eat and drink again. And after a day of recovery in the hospital room, there’ll be only one final thing between the family and going home, a baseline MRI after surgery.

So while the neurosurgeons were confident that they had gotten the whole tumor, tumors can be sneaky bastards and parts can hide only to bloom later. To mitigate against this risk, post-brain surgery patients often get regular follow-up imaging at regular intervals. Because of the risks of ionizing radiation exposure, like from a CT scan, for example, particularly in younger patients, they’re going to want MRI exams. MRI is also often the preferred modality for many types of brain imaging, even if the radiation wasn’t a concern. In order to be able to compare apples to apples, the neurosurgeons want an MRI immediately after the surgery, against which they’ll be comparing future images.

The morning of the day that Michael was set to get discharged from the hospital, he was near the end of his ability to cope with anything that had happened to him over the last few days. He wanted more than anything to just go home. He’d fallen. He’d been examined by scores of doctors and nurses. Had his head shaved. He’d been starved. He’d had his skull cut open. All in all, he was a hurting kid. He looked like he’d been run over by a truck. He just wanted to go home and be surrounded by family and familiar comforts. But the doctors and nurses were insisting on just one more thing, an MRI, then he can go home.

That morning, Michael’s father accompanied his son down to the MRI department. The MRI department was actually an addition built onto the side of the hospital. Mr. Colombini was directed to wait in the lobby of that MRI addition building while they took Michael back for his MRI. In the back of the MRI addition, the technologist was checking Michael in when the anesthesiologist arrived. A face that the MRI tech who had honestly only been there for a few months, but she had never seen that anesthesiologist before. There were some quick hellos between the tech and the anesthesiologist before attention turned back to the agitated boy. Michael still had his IV in from yesterday’s surgery, and the anesthesiologist gave the boy some Versed, a sedative to help calm him down. The medication worked, and Michael was relaxed quickly as the medicine took effect.

The tech finished checking in the boy and wheeled him on a stretcher from the little alcove across the hall into the MRI scanner room. With the anesthesiologist’s help, the two of them slid Michael from the gurney onto the MRI scanner table. That jostling agitated Michael a little bit. He was trying to push through the prior dose of sedative, so the anesthesiologist gave him a booster dose of the sedative, which again, calmed him down. The anesthesiologist placed a nasal cannula on the boy, essentially a plastic tube that wraps around behind the patient’s ears, and can blow supplemental oxygen into the patient’s nose. Then the technologist slides the boy up into the back half of the head coil, which is just a specialized antenna or receiver necessary for getting the MRI images. Then the tech starts bringing the cage looking top half of the head coil down over Michael’s face, and he again pushes through the two administered doses of sedatives. The anesthesiologist gives Michael a third dose of sedative, and now the boy is boneless for his MRI exam. He’s completely compliant and very relaxed.

With the boy correctly positioned, the technologist leaves the MRI scanner room to go back to the control area in order to get things ready to start the actual exam. The anesthesiologist remained in the MRI scanner room, putting a pulse oximetry monitor on the finger of the boy to keep track of how well Michael was breathing and make sure that everything was ready for the MRI exam to start. A pulse oximeter, or pulse ox, measures the percentage of possible oxygen in the blood, with 100 being the top value, 100% oxygenation. When you get below 80 to 85%, you start starving your brain of the oxygen it needs to function the way we want it to. Dizziness and disorientation can start setting in in that range, the 80 to 85 range. Now, below 75, and you’re likely to pass out. Just about as quickly as the anesthesiologist got the finger probe on Michael and got the machine turned on, but he saw the numbers dropping, 97, 96, 95.

This is the precise situation for which anesthesiologists put nasal cannulas on sedated patients. Sometimes too much sedative can slow the patient’s breathing to a point where they’re not quite getting enough oxygen, but if you increase the amount of oxygen they get with each breath, even at slower breathing rates, the patient will still have a healthy oxygen level. So the anesthesiologist goes over to the wall outlet in the MRI scanner room that delivers oxygen into the tube that goes to Michael. When in use, these oxygen outlets have a vertical clear vial, like a glass-like tube, with a neon green ball in it. The higher the rate of flow of oxygen, the higher in the vial the green ball floats. If you want more oxygen, you simply turn the valve until the green floaty ball is hovering at the highest height that corresponds with how much oxygen you want the patient to receive. One liter per minute, two liters per minute, so on.

So all the anesthesiologist needs to do is turn the valve to start the flow of oxygen. Except in this case, turning the valve appears to have no effect. Clockwise, counterclockwise, it makes no difference. That little green floaty ball, yeah, that’s stuck at the bottom of the vial. There is no oxygen being delivered to the boy, and the pulse oximeter reading is still dropping. The anesthesiologist bangs on the window to the control room to get the attention of the MRI tech and points to the door to the MRI scan room, gesturing to the tech to meet him there. When they get there, the tech asks, what do you need, doc? To which the anesthesiologist responds, the oxygen outlet isn’t working, the patient is de-satting, I need you to get the oxygen working.

Here’s where the fact that this building was an addition to the hospital comes into play. It’s in the building code that oxygen outlets are supposed to be piped in the hospital’s central systems with pressure and flow alarms. When there’s a problem, alarms alert engineering teams who can intervene before the systems crash. But this building didn’t have its oxygen piped in from the main hospital system. There was a standalone tank on the other side of the wall with a pipe connecting the tank to the valve in the MRI scanner room. When the tank got emptied, there were no alarms, no alerts, no way to tell the tank was empty until you weren’t getting oxygen out of it. And to get the oxygen flowing again into the MRI scanner room, someone needed to go into the MRI equipment room and manually switch the tanks feeding that wall outlet.

As we said earlier, this MRI tech had only been working at the MRI facility for a few months. She knew that there was something odd with the oxygen system, but she wasn’t exactly sure what that was, or what the appropriate response was. Fortunately, that day she was working with another tech who had been there much longer than her. So she went to go ask him how to fix the oxygen supply to the MRI scanner room. The senior tech said that this was something that she should really know, so he would take her into the MRI equipment room and show her how to switch the tanks. The two of them went into the MRI equipment room without telling the anesthesiologist what they were doing. As soon as the door closed behind them, they were effectively in a soundproof room and could no longer hear anything else that was happening in the rest of the MRI suite.

Now, while all this is happening, Michael’s pulse ox is still dropping 86, 85, 84 and the anesthesiologist is getting increasingly upset, going back and forth between checking the numbers on the pulse oximeter and going for the MRI scanner room door, yelling out to the MRI techs for a status update. But at this point, they can’t hear anything from outside the equipment room. The fact that he’s getting no response is escalating his frustration and anxiety. He’s really getting worried.

It’s at about this time that a nurse who had accompanied a patient down to MRI earlier that day, she returned to the MRI building to get something she’d left behind. We have to imagine that the anesthesiologists yelling carried out into the waiting area where Mr. Colombini was waiting, but that nurse let herself into the back of the MRI addition where the patient care area was, followed that shouting voice back to the MRI scanner room doorway. There she’s able to clearly hear that the anesthesiologist is watching the Colombini boy desaturate, watching his pulse ox values drop. And that anesthesiologist is desperate to get supplemental oxygen to the boy.

This nurse, when she had been down to MRI earlier in the day, had noticed a few portable oxygen cylinders in the little patient preparation alcove directly across the hall from the MRI scan room entrance. They were kind of tucked away, so she wasn’t surprised that the anesthesiologist hadn’t seen them. The anesthesiologist was standing in the doorway, eyes darting back and forth between the pulse oximeter, 81, 80, 79, to the nurse, who was searching for something. Then, miraculously, he saw her pick up an oxygen tank that was about two feet tall, come to the doorway to meet him, and handed it to him to take in to connect to Michael’s nasal cannula.

For those of you who have been with us through all of Series 1 and the telling of the Long Island incident, you already know that how much an object weighs doesn’t tell us much about how much magnetic force it’ll exert when you get it close to the MRI scanner. In Series 1, we speculated that the 20-pound chain around Mr. McAllister’s neck was exerting a thousand pounds of force on his neck. One of the things that increases attractive force is how long a solid magnetic object is. For a chain, that’s probably the length of an individual link of the chain, times however many links there are. But for a steel oxygen tank, that’s the length of the whole tank. Even though an E-size steel tank weighs something similar to the chain around Mr. McAllister’s neck would weigh, the fact that it’s longer means that it can be pulled with significantly more force than a chain could be.

So we have the anesthesiologist who had just been handed this steel oxygen tank, and he’s desperately concerned about Michael Colombini’s oxygen levels. He wants to get this portable tank hooked up to the nasal cannula and quickly. He takes a couple steps into the MRI scanner room towards Michael. We imagine there was a moment, just a fraction of a second, where the anesthesiologist felt the magnet pulling on the tank. In that instant, as the tank was pulling out of his hands, did he realize that a catastrophic accident was happening. 

My patient is de-satting, and the inner oxygen isn’t working. My patient needs oxygen.

What did I see, the portable oxygen cylinders?

Oh, yes.

Hold on, Doc. Here’s an O2 tank.

Thank you. [CRASH]

Next episode, we’re going to take a look at the near-term aftermath through the next several weeks immediately following this tragic accident. So, make sure that you’re subscribed to Invisible Force podcast to be able to get the next episode as soon as it drops. 

For this week’s show, our sources were the court archives of the Colombini family’s lawsuit that named the hospital, the directly involved employees, as well as GE, the MRI scanner manufacturer. 

Our introductory radio play is a dramatization of that critical moment inside Westchester Medical Center’s MRI facility. Our voice actors for this episode’s dramatization are Amanda McLaughlin and Misha Stanton. 

If you have any information about this incident or any other MRI accidents, please reach out to us through our website invisibleforcepodcast.com. Episodes, show notes and a tip line are always available through the website. Also, you can leave us a voicemail with information about MRI incidents or accidents at area code 631-MRI-TIPS. That number again is 631-MRI-TIPS or 631-674-8477. 

And lastly, we like to ask you to like and share our podcast with your friends, colleagues and coworkers. With your help and with a little bit of luck, we’ll help make sure that accidents like this don’t ever occur again.