FYI.

This story is over 5 years old.

Tech

Pigeons Could Help Detect Breast Cancer

Paging Dr. Pigeon!
Image: Levenson et al.

Animals are constantly surprising us when it comes to science—just look at the dogs who can sniff out testicular cancer or the giant rats helping disarm land mines in Mozambique.

The latest species to add to that list is the common pigeon (Columba livia). According to a study by a team of researchers in the US published in PLOS One, pigeons have an amazing ability to generalize patterns—something that could have serious future consequences in medical research.

Advertisement

In a series of experiments, a team comprising researchers from UC Davis, the University of Iowa, and Emory University placed pigeons in front of images of benign and malignant breast cancer cells and rewarded them with food when they pressed a button correctly identifying each of them. Over time, the pigeons became so adept at recognising the cells' properties that when the researchers changed the images to a never-before-seen set, they found the pigeons were able to generalize the patterns from one set to the other and distinguish benign cells from malignant cells.

Over the course of approximately two weeks, they found they were able to train pigeons from 50 percent accuracy up to 85 percent—the same level of accuracy recorded in trained medical professionals.

So what makes pigeons such natural pathologists? "There are predators in the wild that are very well-camouflaged, so it's in the pigeon's best interest to be able to detect the patterns that might reveal if there's any danger in the bushes or not," author Richard Levenson of UC Davis told Motherboard.

The same faculties that allow them to detect predators allow the pigeons to recognise the difference between cancer cells. "Those problems [of evading predators] are mapped quite well to the problems of distinguishing between different kinds of medical images," Levenson added.

"Our results suggest that pigeons can be used as suitable surrogates for human observers."

So what are the implications of these results? In the words of the authors, "Overall, our results suggest that pigeons can be used as suitable surrogates for human observers in certain medical image perception studies, thus avoiding the need to recruit, pay, and retain clinicians as subjects for relatively mundane tasks."

Does that mean a future of battery-farmed pigeon pathologists could be just around the corner? Levenson is hesitant to suggest anything quite so sensational just yet. After all, the role of radiologists is much more than just reading images for signs of cancer.

For now, slightly smaller steps are on the radar. With regard to future research, Levenson said, "We are in discussion with some of my colleagues in artificial intelligence and pathology, and there are some harder data sets that I think we are going to attempt."

We might have to start being a bit nicer to our birdbrain friends.