When Virtual Humans Breathe

(left) Edward E. Hood Endowed Chair Professor of Engineering, George Xu with Department Head, Professor and Director, CeMSIM, Suvranu De
One problem with radiation treatment involves an inescapable fact: patients have to breathe.

Breathing, of course, expands the lungs, and that lung movement moves (or, in the language of the field, “deforms”) all surrounding tissues as well—including tumors. That in turn makes it a stiff challenge to pinpoint the precise location of the tumor at the precise time when the radiation is delivered. Too often, the radiation dose misses the tumor, and the treatment is ineffective.

To make matters more difficult, experts in radiation oncology thought that the solution to this problem—involving 4D imaging technology—could be a black box unless we know how such deformation occurs in reality .

So who better to investigate potential breakthroughs in 4D simulation of the respiratory motion than the researcher who pioneered 3D modeling of human anatomy?

“In this project, the collaboration between nuclear engineers and mechanical engineers was quite easy because we are on the same campus—even in the same department (the Department of Mechanical, Aerospace, and Nuclear Engineering, or MANE, of which De is the head)—and we have a strong tradition of interdisciplinary collaboration.”

Over the past 10-plus years, Rensselaer professor and nuclear engineer George Xu has developed a range of “virtual humans”—computational 3D anatomical models that  accurately portray patient anatomy and physiology. Xu’s earliest fame is probably VIPMan (Visible Photographic Man), an advanced computer model from Visible Human images that simulates in 3D how radiation affects the organs and tissues in the human body. His lab has also created such models as RPI Pregnant Women and RPI Adult Male and Female, as well as RPI Obese Patients, many of which won the best paper awards from the journal of Physics in Medicine and Biology which published their works

As much as these models have contributed to radiation dosimetry, they all have one significant limitation: they are rigid, with constant and fixed geometries for the organs and tissues. Yet organs and tissues rarely stay put—most notably when patients breathe.

That did not stop another Rensselaer professor, Suvranu De, from using tools Xu has developed to investigate surgical simulation techniques, which would enable surgeons to practice complex procedures in a virtual environment before performing them on patients. De, a mechanical engineer, used biomechanical models to simulate tissue deformation in a way the 3D images could not.

That brought De and Xu back to the idea of 4D—and the power of collaboration to make it happen. As a result, the two professors were awarded a $2.2 million grant from the National Institutes of Health (NIH) to create the first 4D virtual human model to simulate breathing and its precise effects on tissue deformation.

The success of the collaboration, from Xu’s perspective, came from a culture of collaboration. “In this project, the collaboration between nuclear engineers and mechanical engineers was quite natural because we are on the same campus—even in the same department (the Department of Mechanical, Aerospace, and Nuclear Engineering, or MANE, of which De is the head)—and we have a strong tradition of interdisciplinary collaboration. Actually, the tradition of interdisciplinary collaboration plays a far more important role than physical location. As engineers we are respected for the engineering expertise and the collaborative problemsolving skills that we bring to a project.”

The collaboration quickly spread to other medical centers including The University of Texas at San Antonio, where a team of radiation oncologists collaborated with Xu to demonstrate the concept of predictive 4D treatment planning for lung cancer patients (the state of the art in the field was based a passive 4D method that could suffer from artifacts). In a related NIH grant, Xu and Peter Caracappa, director of the Reactor Critical Facility in MANE, applied the simulation technology to the protection of nuclear workers, resulting in a series of virtual human models to simulate walking, sitting, and obese employees.

The pioneering research has drawn considerable attention. The team has published nine journal papers and 44 conference abstracts during the 4 years when the projects were funded by the National Institutes of Medicine. Xu and De have been invited to describe their research in plenary and keynote presentations around the world. PhD students and postdoctoral researchers worked on th projects have since been recruited by top companies and medical centers.

Since that time, while continuing to study the generation of geometric models, Xu has been investigating general-purpose graphics processing units, or GPUs, to accelerate computation of the large-size medical images that simulate radiation in the human body. In that new NIH-funded project, he is collaborating with colleagues who are parallel computing experts.

Not surprisingly, Xu continues to stress the central role of collaboration within and across Rensselaer’s low walls. “For Rensselaer to compete effectively in today’s education and research environments, we must continue to foster interdisciplinary approaches on our campus between academic programs,” Xu said. “At the same time, we must also reach out to strategic partners who have complementary research resources.”