Artificial Intelligence for Cancer

Computer programs that mimic human intelligence could be the next new method of colon cancer detection, say researchers.

Researchers from the University of Maryland Greenebaum Cancer Center have devised a method that uses “artificial neural networks,” or ANNs, to analyze thousands of genes simultaneously. ANNs are multi-layer mathematical programs capable of recognizing complex patterns in large amounts of data. Stephen J. Meltzer, M.D., from the University of Maryland, says the technique may one day enable doctors to identify cancer earlier and protect patients from unnecessary surgery.

In a recent study, researchers took nearly 40 tissue samples from patients who had documented cases of cancers related to inflammatory bowel disease. They used an ANN to analyze more than 8,000 genes in each tissue sample. The device was first “trained” to recognize the two types of colon cancer. Researchers then gave it information from 12 samples it had never seen. The ANN made the correct diagnosis in all 12 cases.

Dr. Meltzer is quoted as saying: “This study helps to establish a new method … that can be used in a wide variety of disease settings, not just in cancer. These networks mimic the human brain, in that they can be trained to recognize specific disease lesions or subtle differences within disease categories. Ultimately, we hope that ANNs will greatly aid in the diagnosis and classification of human disease states.”

Researchers say they were able to reduce the number of genes needed for the ANN to make a correct diagnosis from more than 8,000 to 97. They say this will make the method easier and less expensive in the future.

SOURCE: Gastroenterology, 2002