Backing his four-by-four into his garage one night in 2006, after returning from a drunken party in the Swiss canton of Bern, a man crushed his wife against the garage's rear wall.
He called an ambulance, but she died before it reached the hospital. The man claimed it was a tragic accident – he had reversed once and hadn't seen what he was doing – and to the police it looked like a case of involuntary manslaughter. But just to be sure, they asked Michael Thali, director of the University of Bern's Institute of Forensic Medicine, for a second opinion.
Prof Thali and his team carried out an autopsy on the woman, but not the kind we're used to seeing in television shows such as CSI and countless police dramas. This was a scalpel-free, virtual autopsy, or "virtopsy" – a radical new approach to forensic investigation, in which Professor Thali is one of the pioneers.
Using computerised tomography (CT) and magnetic resonance imaging (MRI), the Bern team created a high resolution, 3D virtual double of the woman's crushed corpse. They also scanned the surface of the car so that they could create a virtual model of it, dents and all. Combining these with evidence found at the scene, including skid marks on the garage floor, flakes of paint knocked off the wall and fragments from the car's smashed rear lights, they painstakingly reconstructed the events of that night.
When their virtual reconstruction was complete, they were able to tell the police that the man must have reversed twice, because a single backing up could not account for all the damage they had seen. The first time, they concluded, the woman was standing in a doorway and sustained injuries to her arms and legs, before slumping onto a stool. The second impact damaged her internal organs, and it was those injuries that killed her. The man was charged with wilful homicide. "Ironically, he died before the trial could take place, of natural causes," says Professor Thali. "We performed the autopsy here."
The seeds of the virtopsy concept were sown in the mid-1990s, in a conversation on an airplane between Thali's predecessor, Richard Dirnhofer, and a senior police scientist. They were discussing a high-profile Swiss case which turned on whether a ratchet wrench had caused injuries to the skull of a murder victim. It occurred to them that being able to image those injuries in three dimensions would have been extremely helpful in matching the wound to the weapon that caused them.
What started modestly as a plan to visualise head trauma quickly evolved into a project whose goal was full-body, non-invasive, post mortem dissection. Other forensic pathology laboratories around the world have introduced imaging techniques into their autopsy procedures, but the Bern operation – which performed its 100th virtopsy last year – is by far the most advanced.
Thali and his colleagues' rationale in launching the project was that a scalpel-free autopsy would not only ease the distress of the relatives of the deceased, it would also reduce human error on the part of pathologists and allow them to create a permanent, 3D record of the intact cadaver – something not possible with traditional autopsy, which inevitably destroys tissue and can only be recorded, photographically, in two dimensions. So how does it work? In its first incarnation, the virtopsy procedure involved taking a 3D surface scan of the body using stereoscopic cameras and a projector that cast a stripe pattern onto the body. The team would then slide the table bearing the body into a CT scanner, which would assemble X-ray slices of it into 3D images of the internal organs. Markers placed strategically on the skin during the surface scan allowed surface and interior to be matched.
Later, more high-tech tools were added. While CT is good for revealing bone, it isn't so good for seeing soft tissue, so now they use MRI to image the latter. Taking automation to the next level, they added a robot, the "Virtobot", that can be fitted with a needle and manoeuvred into position by a remote computer, to take a tissue sample should the pathologist need more information. Normally, because a pathologist is guided by CT scans when positioning the biopsy needle, he or she is exposed to X-rays, and even potentially to viruses or toxins contaminating the cadaver. But Virtobot is immune to those. The team can also perform CT angiography, injecting a contrast agent into the corpse's blood vessels and then imaging them to reveal leaks or lesions that might have contributed to death.
At the moment, a virtopsy takes slightly longer than an autopsy, which in straightforward cases can be completed in 90 minutes. But that extra time is compensated for by the quality of the information that a virtopsy provides, and which it can do so non-invasively, says Thali. "Using all this technology together, we see 80 to 90 per cent of forensically relevant findings," he says, adding that a traditional autopsy doesn't reveal 100 per cent of those findings either. Toxicology tests, for example, have to be done separately.
It hasn't all been plain sailing. Professors Thali and Dirnhofer – who coined the term virtopsy – initially encountered a lot of scepticism from the medical community. They are still the only forensic pathology team in the world to have embraced imaging so wholeheartedly. But gradually, says Thali, they are bringing people round. Pathologist Stephan Bolliger was a sceptic, until Professor Dirnhofer showed him what the system could do, and soon after that he joined the virtopsy team. He changed his mind, he says, when he realised that things were possible with a virtopsy that were impossible or very difficult with a traditional autopsy.
For example, he says, "Imagine trying to piece together a couple of dozen fragments of a shattered skull" – something that a pathologist might do to determine the direction and force of a fatal blow to the head – "it could take you days. Now we can do it with the flick of a switch, and we can even show the assembly in 3D to a court." Virtual images are also less distressing to court members than the gruesome photos that pathologists are used to showing.
Another thorn in the side of forensic pathologists is air embolisms. Traumatic injury to the head or neck, for example, can cause an air bubble to enter an artery, travel to the brain and create a blockage there, potentially leading to death. But the air that inevitably enters a corpse with the pathologist's blade can mimic an embolism, making a real one hard to spot. When seen on a scan inside an intact brain, however, an embolism is harder to miss.
Professor Thali's idea is that virtopsies should be used as a filter, with an autopsy only being performed if the virtopsy suggests the cause of death might not have been natural. To date, in Bern, both have been performed as standard in most cases. Occasionally, however, a virtopsy has sufficed. When six young soldiers fell to their deaths while climbing the Jungfrau in 2007, the military district attorney decided that an auto-psy was unnecessary – the cause of death being clear – and asked only for a virtual description of the victims' injuries, so that climbing safety equipment could be improved, and their families informed, hopefully, that they had not suffered.
Virtopsies have already found some more unexpected applications too. In a collaboration with the University of Bern's Institute of Animal Pathology, Thali's team has scanned some of the exotic, expired residents of Swiss zoos, including an aged sun bear that suffered from high blood pressure and died of a stroke, and a young giraffe that strangled itself after getting entangled in a suspended rope basket containing food.
When it is important to preserve the corpse's outward appearance, says radiologist Thomas Ruder of the virtopsy team, the virtual approach is ideal. They have even been asked to virtually "undress" Egyptian and South American mummies, obtaining images of the skeletons without disturbing their funeral robes.
The virtopsy team has high hopes that their techniques will find more mainstream applications. Computer scientist Lars Ebert believes Virtobot may come into its own in the aftermath of major disasters such as the 2004 tsunami when thousands of bodies have to be processed quickly. An on-site Virtobot, plus CT and surface scanners, could potentially be operated by a single technician. who could send the scans via the internet to radiologists overseas.
Radiologist Angela Levy, of Georgetown University Hospital, Washington DC, was involved in research into post-mortem CT scanning for the US military until last year. She agrees that non-invasive scanning has advantages over autopsy, but says there are disadvantages which may restrict its widespread adoption.
"One of the drawbacks for most medical examiner systems is the cost and availability of imaging," she says. The two complement each other very well, she adds, but the use of imaging may realistically be limited to complex or important cases – at least for the time being. The field of imaging is advancing rapidly, though, so watch this space. If and when the cost falls and the machines become more available, virtopsies could be coming to a pathology department near you.