Dog-inspired approach may improve prostate cancer diagnosis, study finds

Feb. 17 (UPI) -- Man's best friend may prove useful diagnosing prostate cancer, serving as the basis for a new screening tool, according to a study published Wednesday by PLOS ONE.

Using their keen sense of smell, dogs correctly identified urine samples collected from men with prostate cancer nearly three-fourth of the time, and correctly identified men who did not have the cancer just as often.

Although using dogs to diagnose cancer in men may not be practical, "artificial neural networks" built based on the canine olfactory system -- their sense of smell -- could be more effective than currently available methods, according to the researchers.

"Machine olfaction inspired by dog noses is coming to a doctor and perhaps even your smartphone soon," study co-author Andreas Mershin told UPI.

"We see the dogs and their training research as teaching our machine learning [sense of smell] and artificial intelligence algorithms how to operate," said Mershin, a physicist at Massachusetts Institute of Technology in Cambridge.

One in eight men in the United States will be diagnosed with prostate cancer during a lifetime, making it the second-most common cancer among men behind melanoma, according to the American Cancer Society.

The prostate specific antigen, or PSA, screening test typically is used to diagnose the disease, but it can miss aggressive prostate cancer in men who have it or indicate that a cancer is aggressive when it really poses little risk, researchers have said.

Two of Mershin's co-authors on the study, Karen S. Sfanos, a pathologist at Johns Hopkins University in Baltimore, and Jonathan Simons, president and CEO of the Prostate Cancer Foundation in Santa Monica, Calif., have devoted much of their work to finding alternatives to PSA testing.

For this study, the researchers trained two dogs to detect aggressive prostate cancer from urine samples collected from 50 men, 12 of whom had biopsy-confirmed prostate cancer.

The dogs, which were trained by Medical Detection Dogs in Milton Keynes, England, correctly identified samples from men with prostate cancer 71% of the time, the researchers said.

In addition, one dog correctly identified men who did not have the disease 76% of the time, while the other was accurate 70% of the time, according to the researchers.

The researchers used the dogs' data to train an artificial neural network to identify cancer cells in collected urine, and the system was able to detect differences between positive and negative samples, they said.

The findings suggest that larger studies could further integrate these methodologies to improve detection of advanced prostate cancer and aid development of new diagnostic tools that replicate dogs' olfactory capabilities, the researchers said.

"Knowing that dogs can be trained to have over 99% accuracy makes us confident [that] replicating this performance by machine olfaction is a useful and possible thing to do for a variety of diagnostics, including prostate cancer," Mershin said.