Closer to a cure

RIT scientist uses math, imaging to help doctors diagnose cancerous tumors

Follow Susan Gawlowicz on Twitter
Follow RITNEWS on Twitter

A. Sue Weisler

Professor Nathan Cahill ’97, ’00 (applied mathematics; industrial and applied mathematics), standing, is improving biomedical image computing, the focus of his Ph.D. research. He and imaging science Ph.D. student Kfir Ben Zikri ’11 (electrical engineering) are developing algorithms for a longitudinal study of lung nodules in CT scans.

Nathan Cahill’s research gives radiologists an edge when analyzing biomedical imagery.

Cahill is a mathematician whose techniques combine imaging science and computational mathematics to enhance the accuracy of images used to diagnose cancerous tumors. His research could someday help radiologists treat lung cancer, the leading cause of cancer deaths in the United States, according to the Center for Disease Control and Prevention.

Cahill, an associate professor in RIT’s School of Mathematical Sciences, and Kfir Ben Zikri, a Ph.D. student in the Chester F. Carlson Center for Imaging Science, are developing algorithms to measure growth in lung nodules—malignant or benign abnormalities—captured by three-dimensional Computed Tomography (CT) scans. Cahill is leading a two-year study, funded by the National Institutes of Health, to compare existing scans taken over time of individual patients with lung cancer.

The rate at which a tumor grows can make a difference in diagnosis, Cahill said. His study will quantify changes in size or show no growth and eliminate the need for biopsies. The algorithm will compensate for complicating background information that blurs edges of nodules.

Simple factors can lead to extraneous background information and introduce error in CT scans. The positioning of the patient during the scan or weight fluctuations between images can stretch or compress the nodules.

The team hopes to win continued funding by extending the algorithm’s capability to isolate and measure nodules even when underlying diseases like emphysema or fibrosis make intensities in the background brighter and the tumor edges harder to see.

Cahill and Ben Zikri’s algorithm will register the background information in a set of images and geometrically transform one image into another. Their approach begins with a mathematical model that describes the physical process. They translate the mathematical model into a programming language that will compute a solution. The researchers run the prototype algorithm on the medical image and interpret the results.

The NIH supports projects like Cahill’s that make technology publicly available. Cahill and Ben Zikri work closely with professor Mark Niethammer, at the University of North Carolina at Chapel Hill, and Kitware, a North Carolina-based open-source software company that specializes in medical image analyses. Kitware will incorporate the algorithm into its free software libraries.

The collaboration grew from a 2011 conference and Cahill’s chance meeting of a likeminded scientist from Kitware.

“This NIH award can be traced directly back to the 2009–2010 RIT Faculty Evaluation and Development award I received,” said Cahill. “It was that award that I used to write a paper and present at the conference.”

Dr. David Fetzer, a radiologist at the University of Pittsburgh Medical Center and a member of the collaboration, suggested the clinical problem. Fetzer, an alumnus from the imaging science program, had worked as an undergraduate with Maria Hegluera, professor in the Center for Imaging Science, and a member of Cahill’s team.

Fetzer is selecting 30 CT scans of patients treated for lung cancer at the University of Pittsburgh Medical Center. The images are scrubbed of patient-identifying information and sent to Cahill and Ben Zikri. Fetzer will clinically verify the algorithmic results that show whether nodules are growing, shrinking or remaining the same size in the anonymous scans.

“This isn’t just an imaging science project, and it isn’t just a math problem,” Cahill said. “It involves doing some complicated mathematical modeling to develop these algorithms that are being used to solve a problem in imaging.”

Cahill’s research interests routinely cross disciplines and colleges. He is a graduate faculty member in the Center for Imaging Science and is on the extended faculty of the Ph.D. program in computing and information.

“Once I started working in the medical area, I liked it because my mother’s a nurse, and I have an uncle who’s a radiologist,” Cahill said. “All of a sudden it clicked. I’m using my skills in developing mathematical models and implementing ideas where I can see the importance. It’s a field that has the potential to help people.”