The Transparent Patient

Though relying on an impressive array of visualization technologies—magnetic resonance (MR), computed tomography (CT), ultrasound and others—many of today's surgical procedures are still surprisingly similar to those of twenty years ago. Surgeons still open their patients with little more to guide them to a tumor, appendix or polyp's location than the images in their heads and the occasional look over the shoulder at a monitor. But things are set to change. Just around the corner, technologies are taking shape that will make today's operations look like the technological equivalent of charting a course with nothing more than a compass and sextant.

But the potential of in situ visualization goes well beyond deciding where to cut and how large an opening should be made. A major problem in surgery is that once the patient is open, organs tend to move. This is especially true—and particularly dangerous—when it comes to brain surgery. Working with Siemens Medical Systems, Rubino has therefore taken the unusual step of having an MR scanner installed in an operating room. The idea is to provide dynamic images of brain anatomy during operations, thus helping to guide neurosurgeons to the exact location of a tumor. To date, he and his team have performed over 100 intra-operative MR (iMR) neurological procedures. During such operations, the patient is periodically swiveled under the MR machine to produce updated images as needed.

The next step is to transfer the images that would normally appear on monitors to augmented reality devices, thus providing a kind of 3D roadmap to the target. Says Frank Sauer, Ph.D. of Siemens Corporate Research (SCR) in Princeton, NJ, who has been developing medical HMD applications and is working closely with Rubino, "What is unique about our technology is that it offers a dynamic viewpoint. Thanks to a 'marker bridge' dotted with optical markers and attached to the frame that holds the patient's head still during the operation, an infrared sensor on the HMD can track the precise distance and angle of the surgeon's head vis-à-vis the patient's head. This allows the surgeon to go around the patient and see the tumor from different angles and helps him to intuitively see the best route to a tumor. It holds the potential for improving outcome and cutting costs."

A range of diagnostic images fed into the surgeon's field of view showing the real time location of tumors and other hidden structures—that's just a peek at how augmented reality devices will support surgery over the next few years. But many other information sources may eventually find their way into this all-purpose instrument, including diagnostic information gleaned from hospital and other data bases. The technologies outlined over the next few pages all fit this model. Some, such as virtual colonoscopy, are purely diagnostic today; but married to advanced in situ visualization devices, they will play an important role in simulating surgical procedures and in helping surgeons to rapidly zero in on their targets.

But virtual colonoscopies still require real prepping—the unpleasant business of clearing the intestine and preparing it for visualization. So Geiger and others are working on a digital fix. "It may be possible to digitally remove the stool from the images," says Geiger. "If we could obviate the prepping, that would remove one of the main reasons for avoiding the test. Besides, it would make it much faster and cheaper."

Before a virtual exam can take the place of a real one, its data have to be transformed into images that are so realistic that from an internist's point of view, the two are identical. That's the job of imaging maestro Gianluca Paladini, 37. Paladini has come up with a "Rendering Engine"—a program that calculates the colors and brightnesses of surfaces based on their geometry, material composition and lighting—for SCR's Imaging and Visualization Toolkit (IVT). The engine produces 3D reconstructions of CT and MR data of unprecedented quality, detail and realism. At the heart of this achievement is an algorithm he developed that supercharges a notoriously slow image computation technique called 'ray casting.' "Ray casting normally takes 30 seconds or more to compute one image," says Paladini. "But by using several optimization techniques, I implemented a ray casting algorithm that computes at an interactive frame rate." The result is that doctors can now zip back and forth through a tract of intestine at 15 frames a second inspecting every nook and cranny that might harbor polyps.

As imaging modalities such as MR and CT are used to produce specialized studies such as virtual colonoscopy, doctors are being confronted with a new level of information overload. "We will soon be up to a gigabyte per patient," say Alok Gupta, Ph.D., who heads SCR's Imaging and Visualization Department. "What's needed is the digital equivalent of a gold miner's pan." With this in mind, Gupta's team is homing in on a new field he calls "interactive computer-aided diagnostics." The idea is to allow radiologists to look at, say, 500 CT images in an acceptable time frame, but without missing crucial diagnostic data. "We are developing techniques that allow doctors to highlight regions that look suspicious, isolate them from the volume data and make measurements," says Gupta. The technique can be particularly useful when it comes to follow-up studies of slowly developing conditions such as pre-cancerous liver nodules where exactly the same nodules must be located and their volumes compared over time.

Shahram Hejazi, Ph.D., is used to focusing on the future. His job, as head of the Health Innovation Field at Siemens Corporate Research, is to look ahead, analyze new markets, and identify tomorrow's business opportunities. When it comes to surgery, what he sees might be summarized as 'smaller is beautiful.' "Twenty years ago knee surgery had a recuperation time of almost three months. Now, it's a week or two. Why? Because instead of cutting the knee wide open, we get the job done with two or three punctures. The tools have gotten smaller and smarter." The same goes for any number of procedures in gastroenterology, neurosurgery, and cardiac surgery. But wait until you see what's on the horizon.

In the case of cancer surgery, the major obstacle to success is that stray cells can be left behind. How can a surgeon see those cells? One likely answer is called 'molecular imaging'—a developing technology that uses smart and sophisticated imaging agents that operate in the near infrared spectrum so that detection is possible with less expensive and more widespread machines than is currently the case. With this in mind, scientists at Massachusetts General Hospital in Boston have developed molecules that glow at infrared wavelengths when exposed to the metabolic processes in cancer cells. Says Peter Kleinschmidt, Vice President for R&D at Siemens Medical Solutions, "we have already developed and tested a device that detects these molecule when they glow. We are onto a new sort of imaging system that identifies cancer cells at a very early stage. The potential of this method is outstanding because it allows specific forms of tumor to be differentiated. We call this the virtual biopsy."

Also under development are molecular structures that, according to Kleinschmidt, become visible for an MR machine when they come into contact with a tumor. "Since these strategies work on a cellular basis, they hold the potential for treating cancers cell by cell with a chemical scalpel while leaving normal tissues completely intact," he says. Treating? Sure. "We are working on superimposing MR images of cancer tissues over real patient anatomy. It's augmented reality of affected organs as visualized by molecular processes," says Hejazi. "And given the right tools, this points in the direction of molecular surgery."

Although the mechanics of how surgeons will eventually cut and suture tissues on a molecular level are still far from clear (indeed, such treatments may never evolve because medical cures may prove far more practical), Hejazi and Kleinschmidt have little doubt that surgical instruments will shrink rapidly over the next few years. "The new thing is minimally invasive surgery," says Hejazi. "But we will soon reach a point at which human hands and eyes can no longer keep up. That's where assisted robotics comes in."

Remotely guided by a surgeon's hands, robotic instruments are already in use in a handful of centers (see Interviews with Experts). In fact, claims Intuitive Surgical Inc, maker of a leading robotic OR system, hundreds of such minimally invasive, robotically assisted procedures ranging from mitral valve repairs to coronary artery grafting, have taken place. "The robotic tools now in use are still relatively clumsy," comments Kleinschmidt. "But my opinion is that in coming years surgeons will be supported to an almost unimaginable degree by new, miniaturized versions of these technologies." The trend toward barely visible robotic instruments will bring with it a number of advantages. Human tremor, for instance, will vanish at the interface between man and machine; sterile areas will not be put at risk by human contact; comprehensive surgical planning—the exact visualization of the paths instruments will take during surgical procedures—will be simplified because the exact dimensions and movements of the instruments are known; anesthesia will be reduced; wounds will be smaller, and recovery will be greatly accelerated.

As extraordinary as the advances on surgery's near-term horizon appear to be, there is no denying the fact that the ancient art of cutting and healing boils down to nothing more than fixing things that have gone wrong. But suppose we could keep ourselves from breaking down in the first place? The key to accomplishing that is called gene and protein testing. "Siemens has realized that this kind of testing is where the future lies," say Kleinschmidt, "and we are working with other companies to bring products to market." Down the road, according to Kleinschmidt, are microchips that will be able to perform thousands of genetic and possibly even protein tests simultaneously. The implications of a protein analysis product are mind-boggling. Your general practitioner will be able to simply take a blood sample, inject it into a sensing device about the size of a telephone answering machine, and receive a report within minutes as to whether you have early signs of illnesses such as prostate cancer, asthma or diabetes. Such devices could work by exposing a blood or sputum sample to antibodies designed to fluoresce if they come into contact with abnormal proteins. Exposed to laser light, the antibodies will then emit light. Abnormal proteins will be identified by the wavelengths they emit. "This technology will allow us to discover nascent tumors years before it would be possible to see them using imaging systems," says Kleinschmidt.

Arthur F. Pease