He has created a variety of instruments that can travel the esophagus, stomach and intestine, allowing pathologists to probe for cancers by peering in three dimensions below the surface of the skin.
Frustrated by the time between when a tissue sample is taken and when a pathology laboratory can examine it, Dr. Contag, who oversees a molecular imaging laboratory at Stanford, is experimenting with a variety of next-generation endoscopes. The new devices not only portray the surface of the skin, but also use a variety of optical and acoustical techniques to virtually “punch holes” in hundreds of cells deep within the human body, while using contrast agents to identify abnormalities. He describes the approach as “point-of-care pathology,” part of a convergence of medical technologies that make it increasingly possible for surgeons and medical technicians to make informed, on-the-spot decisions about patient care. “We want to give the pathologist what he already looks at, so it’s pretty easy,” he said. In the half-century since the movie “Fantastic Voyage” portrayed a miniaturized submarine navigating the human body to find and destroy a blood clot, researchers have relied on optical, magnetic and X-ray imaging techniques to peer into bodies with ever-greater precision.
Today a new wave of imaging technologies is again transforming the practice of medicine. They include new pathology tools — like the ones Dr. Contag’s team is developing — to give doctors an instantaneous diagnosis, as well as inexpensive systems, often based on smartphones, that can extend advanced imaging technologies to the entire world. On the horizon is magnetic imaging technology that will combine the speed of X-ray-based computerized tomography, or CT, with the ability of M.R.I. systems to image soft tissues.
The advances are being driven largely by the falling cost of computing, as well as the increasing availability of other miniaturization technologies, including nanotechnology. Dr. Contag said he faces challenges, especially from the medical community, which still has to be convinced that computerized images can equal the precision of laboratory practices in which a pathologist conducts a range of tests to determine whether a specimen has healthy or diseased tissue. But that may change soon. Dr. Contag is pursuing a new generation of molecular biomarkers that can be injected and then attach to lesions, giving doctors a direct answer about disease on a cell-by-cell basis. “You don’t need machine learning, you don’t need machine vision,” he said. On a computer screen he showed an image of a digital sample, with areas that were distinctively brighter. “That’s cancer; that’s normal,” he said, pointing to the dark and light sections. Advances in digital imaging are also transforming conventional laboratory tools. In a lab at Columbia University Medical Center, Matthew Putman shows how software can speed the work of a human pathologist. Dr. Putman specializes in the design of advanced polymers. However, his research requires advanced imaging software, and that has led to the development of new computerized analysis tools.
“You will still see some labs in this building that use manual inspection,” he said. In a neurology lab here, for example, slices of a mouse brain a single cell thick are placed on microscope slides. “People here bend over microscopes, find the hippocampus and take images of it,” he said. “To do one slice can take a day.” In contrast, his nSPEC pattern recognition software can automatically scan 12 slides and generate the same results in just 15 minutes, he said. The software can be trained to identify a wide variety of biological structures ranging from neurons in the brain to pathogens. Dr. Putman’s firm, Nanotronics Imaging, in Cuyahoga Falls, Ohio, is collaborating with Jamaica Hospital Medical Center in Queens in a trial study using the software to automatically identify squamous cell abnormalities typically found in a Pap test. Typically the test requires that a laboratory technician examine at least 5,000 cells. “Our idea is to bring it in to the patient,” he said. “You automatically run the Pap test — you run it, you screen it and get results right there.” In this case the system will not replace the pathologist, but it will replace the centralized multimillion-dollar systems that now do the preliminary screening looking for abnormalities. Other traditional imaging technologies are being rapidly transformed by computation. For example, similar to Dr. Contag’s research with endoscopes, the electronics corporation Philips has developed an advanced ultrasound system that is inserted through a patient’s mouth into the esophagus.
Known as three-dimensional transesophageal echocardiography, or 3-D TEE, the technique produces an image of the heart from inside the patient’s rib cage, which often prevents ultrasound from capturing clear images. Computer processing of the data, which is transported by a fiber-optic cable from the sensor, creates stunning high-resolution 3-D videos of beating hearts. More recently, Philips has used computation extensively in its Heart Navigator system, which provides a three-dimensional map for a cardiologist. Only recently certified in the United States by the Food and Drug Administration, it has made the implantation of replacement heart valves by catheter routine in Europe. “The patient can walk out of the hospital immediately after being treated,” said Bert van Meurs, senior vice president and general manager of Philips’ Interventional X-Ray group. Philips is also developing an imaging technology called magnetic particle imaging, or M.P.I. It requires the injection of a magnetic “tracer,” but the result is higher resolution and faster imaging. Consequently, M.P.I. could play a crucial role in surgical procedures that require simultaneous imaging and now use technologies like CT and PET scans. The system holds promise both for tumor diagnosis and coronary assessments, said Joern Borgert, senior scientist at Philips Research in Hamburg, Germany, and leader of the M.P.I. development project.
Source: The New York Times
