Profile

The 3-D world of Geoff Rubin
Computer-enhanced imaging's evangelist maps radiology's future

By Jane Lowers

Amid reconstructions of the aorta and cranium on the wall of Stanford University's 3-D imaging lab is a virtual cystoscopy revealing a tumor in a patient's bladder. Imaging researcher Dr. Geoff Rubin downplays it--it's eye candy compared with the detailed aortic aneurysm measurements his staff produces daily. But it has potential.

"Right now, because 3-D rendering is a labor-intensive step, we focus only on the work that really adds clinical value," he said. "Someday, 3-D, 4-D, even 5-D imaging with time variation and multi-acquisition fusion may be the diagnostic standard."

Even as he serves as codirector for the 3-D center, which turns out up to 275 vascular and musculoskeletal images a month, Rubin envisions its evolution. Backed by formidable computing power, radiologists and referring physicians could work from such detailed imaging on their own workstations. Software algorithms would produce the measurements now painstakingly completed by a staff of technologists and engineers.

The potential excites Rubin, who first glimpsed his own future when Stanford acquired one of the nation's first spiral CT scanners in 1991. Returning from the RSNA meeting, where he'd heard much about the machine's capabilities, the third-year resident asked for permission to scan a patient's renal arteries during contrast injection prior to brain imaging for a meningioma. The resulting images sparked acute interest in the department and provided a focus for Rubin's career.

Ten years later, the affable Rubin treads the same hallways as an associate professor, dividing his time between clinical and academic work. Though his research has spanned several modalities and a range of anatomy, today he and collaborator Sandy Napel, Ph.D., an electrical engineer, focus on CT's ability to measure the morphology and function of the cardiovascular system and on identifying pulmonary nodules via computer-aided detection.

A portion of Rubin's research looks further ahead, reinventing the way radiologists interact with data--to the point of sweeping away monitors and keyboards in favor of touchscreens and even feedback-enhanced gloves. Facing a rotating 3-D model of a patient's abdomen, a viewer would be able to strip away undesired muscle, organs, and bone with the flick of a finger, revealing an aortic aneurysm and its surrounding arteries. Closer inspection would reveal points of weakness and tension along its walls, the location and volume of a thrombus in the sac, and the best location in the neck to secure a stent-graft. Real-time mapping would guide an interventionalist through the placement procedure. The computer becomes a silent servant, anticipating needs and providing intuitive options.

"A technologist in our 3-D lab used to spend hours generating a curved planar reformation with blood vessels; very labor-intensive work," Napel said. "We now have the software to do it automatically. What's great about Geoff is that he has both the scientific and medical backgrounds to understand what physicians need and what kind of algorithms we need to create to get there."

For a man whose work is tied so closely to computers, Rubin came to them relatively late, buying a Macintosh in his fourth year of residency--to do his taxes. He pursued a science career initially as a way to support his music habit. An aspiring electric bass player through his undergraduate years at the California Institute of Technology, he spent a year playing big band, new wave, and jazz before entering medical school at the University of California, San Diego. He still plays almost daily, with his six-year-old son, who recently started piano lessons. Rubin hopes that a father's enthusiasm will also take root in his other children: triplet kindergartners and a toddler.

"When I have moments of inspiration, if you can call it that, about my research, invariably it's when I'm playing music or with my kids," he said. "Nothing parallels the experience of feeling that communication with other players that transcends spoken word."

If musical collaboration is hard to come by in the life of an academic radiologist, research collaboration isn't. With radiologist R. Brooke Jeffrey, Napel, and other Stanford colleagues, Rubin has contributed to more than 60 papers in the past eight years.

"Geoff is doing remarkable, very focused, and clinically relevant work here. What's more, he takes it on the road and gets other people excited about the possibilities of things like CT angiography," Jeffrey said. "He's a luminary, and you can't say that about many people. "

It's the potential-laden atmosphere of radiology's growth spurt that has captured Rubin's imagination. Although he's proud of the exquisite detail his 3-D lab can produce on paper for clinicians to show their patients, he can't help but imagine that the data could be more useful if computers could be used routinely to perform the manipulations radiologists must now approximate in their heads.

It won't all have fused functional and morphological information, and it won't all contain timed flow sequences--but it could, if that's what is most useful for the case in question.

"I see us moving toward a period where we don't look at a stack of cross sections unless that's what we want to see, and the plane of those cross sections is arbitrarily defined based on the question at hand," Rubin said. "We should interact with the data in an effective and efficient manner, where efficiency is defined not only by how fast you are but by how accurate."