This report originally appeared in the June 2011 issue of DOTmed Business News

The Radiological Society of North America’s annual tradeshow is the world’s biggest radiologist convention and one of the largest medical conferences around. But it’s not the sort of place you’d expect to see autopsy photos.

Yet, last year, attendees saw images from an autopsy – at least, a virtual one.

On the show floor, the Swedish diagnostic imaging company Sectra was demonstrating its “virtual autopsy” table – a touch-screen display hooked up to a PACS system. About the size of a small pool table, it’s what one might expect to see on an episode of CSI.

At the show, the table’s screen showed images of a rather flabby, middle-aged man, who was apparently stabbed to death in his apartment. After he was found, the medical examiner put him through a CT scanner, creating a three-dimensional X-ray portrait of his corpse.

With the CT reconstruction of the body, doctors could then examine the cadaver inside and out.

First, they could look at the skin surface, and see the small, straw-hole gashes left by the knife. Peeling away the skin and viscera, they could examine the ribs smashed by the force of the attack. And they could also find something not revealed during a normal autopsy: the pockets of air trapped inside body, formed where the flesh closed around the blade.

Although the table is brand-new, virtual autopsies, technically, are not. CT scans on corpses have been going on since at least the 1990s.

But only in the last decade has a formal process arisen for managing caseloads. The leader in this field, a Swiss group, calls that process “virtopsy.”

“We’re not the first to perform CT scans on a dead body,” Dr. Lars Ebert, a computer scientist with the Virtopsy group, based out of the University of Zurich, told DOTmed News. “But we’re the first to generate a workflow for it, and to implement it as a standard procedure, and make a scientific approach with bigger numbers of cases.”

Autopsies go virtual
The roots of the Virtopsy project stretch back to 2000.

Around that time, researchers made a surface scan of a child run over by a car, to match the impressions on the skin with tire tracks from the vehicle. This was among a handful of cases that prompted the researchers to wonder if the process could move from an occasional, one-off tool to a “black box system” for routine autopsies.

“You put in body on one side, it comes out other side, and the system gives you an idea about the injuries and documents everything,” Ebert said.

Virtopsy, in essence, was born. By 2006, the team, then based in Bern, got its own CT scanner. Later, the researchers also picked up a surface scanner and a 1.5-T MRI unit. By the time they moved to Zurich, they had been running around 300-400 cases a year (and they also upgraded to a 3T MRI). In Zurich, they expect to do slightly more – though not, perhaps, much more.




“Switzerland is not the U.S.,” Ebert wryly noted.

Helpful, but expensive
The main advantage of the virtopsy approach is its comprehensiveness, Ebert said. A pathologist dissecting a cadaver takes notes about what he sees, but he doesn’t write down everything. If you take a CT or MRI scan, you have the complete body regardless of what you’re looking for, Ebert said. If new questions arise, you can always go back to re-examine the slides.

Yet Ebert cautioned that a virtual autopsy is not a substitute for a normal autopsy, as medical examiners cannot identify all pathologies and diseases just using MRI and CT scans.

“A CT scan can be a good adjunct to an autopsy,” Dr. Stephen Cina, deputy chief medical examiner of Broward County, Fla., professor of pathology at the University of Miami and president of the Florida Association of Medical Examiners, told DOTmed News. “But an autopsy is still the gold standard.”



Currently, Cina said the recommended use for CT scans is to help guide the autopsy, by identifying injuries or pathological conditions in areas not routinely dissected by examiners. For instance, examiners often respect family wishes to hold open-casket funerals. This means they rarely cut deep inside the facial bones, so a CT scan could be useful to get preliminary information about the face.

It’s also helpful in high-profile cases where a lot evidence is needed, or if the family has strong religious objections to autopsies, such as among Orthodox Jews.

But with the exception of the military, where nearly all soldiers killed in combat are scanned when their bodies return home, virtual autopsy technology is not routinely used by medical examiners in the United States [See Sidebar]. Fewer than 10 of the big U.S. examiner offices either have the technology or are planning to buy it in the near future, Cina said.

The main problem with adopting the technology is the one afflicting nearly every aspect of health care: cost.

In his recent survey of 100 medical examiners, nearly one-fifth of all practicing, board-certified examiners in the country, Cina said he found the majority would use virtual autopsy technology if it were accessible and affordable.

“The trouble is, you’re also working in a medial examiner system on a county or state budget,” he said. And with ongoing financial crises, it’s hard to ask for funding for CT or MRI machines, which cost millions of dollars.

In fact, the Virtopsy group estimates a full-body CT scan adds about 1,000-1,500 Swiss francs per exam ($1,134 - $1,701). MRI, which generally carries $100,000 in yearly maintenance costs, probably also adds 1,500 Swiss francs.

Another snag for adoption is the need for specialists to interpret the images. In general, the Virtopsy group says doctors with some training in an emerging discipline, forensic radiology, are preferred for image interpretation. “A normal radiologist is not the person you need,” Ebert said. “Some of the changes you see postmortem could be mistaken [for another pathology]. You need to read images differently.”

Hands-free PACS manipulation
Whether it becomes a standard tool in the United States or not, the Swiss team is still working to make virtopsies as efficient as possible. And recently, they found a $140 entertainment device that might give virtopsy examiners’ workflow a boost.

The device is Microsoft’s Kinect, and the problem it solves is a gory one.

The autopsy hall is, after all, a messy place. Medical examiners get their hands dirty with body fluids and other contaminants. If they want to browse through CT slides or look at other imaging data on a PACS system, they have to remove their gloves and wash up, or ask someone to operate the computer for them. This takes time. But the Virtopsy group thinks the Kinect might be able to simplify this.

Released last fall by Microsoft for the Xbox 360 video game console, the Kinect is, in essence, a camera system with a microphone that lets players manipulate on-screen images using just body movements, gestures and voice commands. It was designed to compete with Nintendo’s massively popular Wii, and was launched with a slew of family-friendly titles like “Kinectimals” – where players interact with adorable baby animals.

But many people saw it had potential beyond playing with cartoon tiger cubs. Shortly after its Nov. 4 launch, open-source enthusiasts Adafruit Industries put up a $3,000 bounty on the first person to come up with an open-source driver for the device. By Nov. 10, hacker Hector Martin claimed the prize. Soon, artists, scientists and entrepreneurs were finding new uses for it. One team developed a program so a performer could strum an “air guitar” that actually creates music. Others figured out how to use gestures to control robots and computer-rendered puppets.

At first, Microsoft’s response was, as expected, rather ham-fisted. The company threatened to “work closely with law enforcement” to keep the Kinect “tamper-resistant.” But the Redmond, Wash.-based software giant eventually accepted the Kinect was more than a toy, and decided to embrace the new uses. This spring, Microsoft is even scheduled to release its own basic programming toolkit.

The potential for the Kinect wasn’t lost on the Virtopsy team. Around the time the Kinect hacks made news, Ebert said Dr. Steffen Ross, a gadget-crazy colleague, approached him with an idea. What if the Kinect could be used for the hands-free control of the PACS system in the autopsy hall? This way, the doctor could just wave his arms around to scroll through the slides. No washing up needed.

Ebert, at first, was skeptical. He wasn’t sure if he could get the Kinect and the PACS to work together. “I had no idea how complicated it was to [get] these hack drivers,” he said, so nothing was done. But sometime later, the colleague went out and bought a Kinect anyway. Before leaving for a conference in the United States, he left it in Ebert’s office. Ebert got his hands on an open-source hack driver, and started to tinker around with it. A few weeks later, it was working.

The Kinect isn’t the first time someone has tried to create a no-touch PACS interface. An earlier piece of software, called Gestix, could be set up to remotely operate a PACS system – through gestures – using a Web cam. But Ebert said because it relied on a basic Web cam, it was hard to program. The Web cam would have to be calibrated to recognize a physician’s specific gestures and even the color of the glove used.

The ease with which the team got Kinect working comes from its real strength: it’s a depth-sensing camera. The device is actually a two-part system: a light source shoots out infrared beams to map the room, which get reflected back and picked up by the camera. The system can then tell how far into the room something is by how long it takes for the light to bounce back. It can then separate the foreground – say, a doctor’s hand– from the background. It can use this “knowledge” to only interpret the nearer object’s movements: in this case, the doctor’s hands. Programmers can even ensure the Kinect isn’t confused by someone walking in front of the doctor, by telling it to ignore near, big objects that suddenly appear.

“You have depth, not just color, so it makes it quite easy to program and get results quickly,” Ebert said.

Nonetheless, the team’s Kinect-PACS control system is just a proof-of-concept prototype. It hasn’t been used in a real virtopsy case yet. “Basically, we wanted to publish [the results] first before we continue development. Even though my colleagues already were like, ‘When can we use it?’”

Ebert said they’ve submitted their findings to the journal Surgical Innovation, and are awaiting a response.

But they already know some of the challenges ahead, such as perfecting the device’s voice-control. The system doesn’t pick up Swiss-German accented English very well, the accent of most of the Virtopsy examiners. And the team still would prefer to have finger recognition, which is being worked out by computer experts at MIT.

Naturally, the Virtopsy team isn’t alone. As with the iPad, other doctors are busy trying to find a place for the Kinect in medicine. Doctors at Sunnybrook Hospital in Toronto, for instance, have done something similar to the Virtopsy team, rigging up a Kinect to control a PACS system during surgeries, so doctors can flip through slides without leaving the sterile field. And a team of students has developed a way to control a multimillion-dollar surgical robot – hands-free, of course.

EXTRA:

Scanning the fallen
In the United States, the technology that forms the backbone of the virtual autopsy process has caught on with the Department of Defense. Over the past decade, most fallen soldiers coming back from the wars in the Middle East undergo CT scans in the autopsy suite over in Dover Air Force Base.

In fact, Dr. Michael Thali, managing director of the Institute of Forensic Medicine at the University Zurich and one of the founders of the Virtopsy project in Switzerland, worked at the Armed Forces Institute of Pathology as a researcher earlier in his career.

Paul Stone, an Army spokesman, said the CT scans help the doctors examining the soldiers map bullet trajectories and help them find shrapnel in, say, the back, which might be ignored in a normal autopsy.

“We’re not doing virtual autopsies,” he told DOTmed News. “We’re using the CT scanner to augment or enhance the autopsy process.”

The scans also help the military’s data-gathering efforts to better understand how our soldiers are killed, information that can be useful in protecting them. Because the CT scans can be done on soldiers still kitted out in their full-body armor, researchers can see exactly where the armor failed to save them.

While this information is shared with the teams who develop the armor, Stone suggested it’s too simplistic to say armor improvements come directly from their efforts.

“It’s very complex, the data they use to improve or redesign things,” Stone said. “But we support that effort, we provide data, here and from the folks fighting the wars in Afghanistan and Iraq.”

Products mentioned in this article

Kinect. Made by Microsoft Corp. Retail price: about $140. (http://www.xbox.com/en-US/Kinect/)

Sectra Visualization Table. Made by Sectra Imtec AB. List price: about 160,000 euros. (http://www.sectra.com/medical/visualization/forensics/)