Mine-hunting software helping doctors identify cancer cells

October 7, 2011

A human kidney cancer specimen from the FARSIGHT medical image analysis software is enhanced by an Office of Naval Research-funded active learning algorithm. In green, the software highlights a thin layer of cells of interest. (Credit: U.S. Navy)

Duke University researchers have adapted Office of Naval Research (ONR)-funded software used for finding and recognizing undersea mines to help doctors identify and classify cancer-related cells.

When examining tissue samples, doctors must currently sift through hundreds of microscopic images containing millions of cells. To pinpoint specific cells of interest, they use an automated image analysis software toolkit called FARSIGHT (Fluorescence Association Rules for Quantitative Insight). It identifies cells based upon on examples initially labeled by a physician. But the resulting classifications can be erroneous because the computer applies tags based on a small sampling.

The Duke scientists said they have added the ONR active learning software algorithms to FARSIGHT  to make the identification of cells more accurate and more consistent. The enhanced toolkit also requires physicians to label fewer cell samples because the algorithm automatically selects the best set of examples to teach the software.

How objects in the ocean relate to cancer cells

ONR’s active learning software was originally developed to allow robotic mine-hunting systems to behave more like humans when they are uncertain about how to classify an object. Using information theory, the software asks a human to provide labels for those items. This feature is valuable in mine warfare, where identifying unknown objects beneath the ocean has been accomplished traditionally by sending in divers.

A medical team at the University of Pennsylvania is now applying the ONR algorithms, embedded into FARSIGHT, to examine tumors from kidney cancer patients. Focusing on endothelial cells that form the blood vessels that supply the tumors with oxygen and nutrients, the research could one day improve drug treatments for different types of kidney cancer, also known as renal cell carcinoma.

“With the computer program having learned to pick out an endothelial cell, we have now automated this process, and it seems to be highly accurate,” said Dr. William Lee, an associate professor of medicine, hematology and oncology at the university who is leading the research effort. It usually takes days, even weeks, for a pathologist to manually pick out all the endothelial cells in 100 images. The enhanced FARSIGHT toolkit can accomplish the same feat in a few hours with human accuracy, he said.