Artificial Intelligence Used to Diagnose Colon Tumors

Originally Published MDDI April 2002

R&D DIGEST

Although patients with Crohn's disease and ulcerative colitis, the two forms of inflammatory bowel disease (IBD), have an increased risk of developing cancer, the cancer can be one of two forms. Common colon cancers can often be removed without radical surgery. IBD-related growths and cancers, however, are more aggressive and most often require removal of the entire colon.

"Until now, we had no reliable way to discriminate between these two types of lesions, especially in their early stages," says Stephen J. Meltzer, MD, professor of medicine at the University of Maryland School of Medicine (Baltimore), associate director for core sciences at the University of Maryland Greenebaum Cancer Center, and director of the cancer center's Genomics Core Facility. Researchers at the university are developing highly sophisticated computer programs that use artificial neural networks (ANNs) to analyze thousands of genes simultaneously. Meltzer says the research could ultimately help doctors to identify cancers earlier and prevent unnecessary surgery.

According to Florin M. Selaru, MD, research associate in the Department of Medicine at the University of Maryland School of Medicine and director of bioinformatics and data analysis at the Greenebaum center, the study entailed taking tissue samples from 39 patients at the VA Medical Center (Baltimore) and Mount Sinai Hospital (New York City) who had well-documented cases of "sporadic" cancers or cancers related to IBD. DNA was extracted from the samples and the researchers used microarray equipment to analyze 8064 genes to determine the level at which they were present in each colon sample.

Selaru explains that these "gene expression" levels were translated into numbers, which were processed by the ANNs. Using gene information from 27 of the 39 samples, researchers "trained" the neural network to recognize the two types of colon cancer. When the network was then given information from 12 samples it had never seen, it made the correct diagnosis in all 12 cases.

Copyright ©2002 Medical Device & Diagnostic Industry