Pages

15 December 2015

Pigeon vision: ‘flocksourcing’ cancer detection

Researchers are learning more about how to improve cancer
detection through teaching pigeons like the two above
to identify images of cancerous cells.
Pigeons have been taught how to detect breast cancer -- with an accuracy rate that surpasses humans -- and in the process have inspired ideas about how to better teach humans how to visually detect cancer.

Researchers from the University of California Davis, the University of Iowa, and Emory University published a paper last month detailing how they trained pigeons -- Columba livia, commonly called rock doves, to be precise -- to detect cancerous cells. The birds attained an accuracy rate of 85%, higher than the accuracy of humans doing the task (84%), the Chicago Tribune reported. (Also see the Wall Street Journal for more coverage.)

And when four pigeons were tested on the image and their results combined (“flocksourcing”?), the birds were 99% accurate in identifying cancerous cells.

The researchers also found that while the pigeons had high-accuracy results when looking at slides from tissue samples, they were not able to learn how to accurately identify signs of cancer when looking at mammograms. Unlike biopsied cells viewed under magnification, mammogram images show neighboring tissues such as blood vessels, a factor which affects human accuracy as well.

Because a pigeon’s vision works much the same as a human’s, the research could help scientists improve the results in teaching humans how to visually identify cancer.

“Pathologists and radiologists spend years acquiring and refining their medically essential visual skills, so it is of considerable interest to understand how this process actually unfolds and what image features and properties are critical for accurate diagnostic performance,” the researchers wrote in their article in PLoS ONE.

The research team included Edward Wasserman, Stuit Professor of Experimental Psychology at the University of Iowa; Elizabeth Krupinksi, professor and Vice Chair for Research in the Department of Radiology and Imaging Sciences at Emory University; Richard Levenson, professor and Vice Chair for Strategic Technologies in the Department of Pathology and Laboratory Medicine at the University of California Davis Medical Center; and Victor Navarro, a graduate student in the Department of Psychological and Brain Sciences at the University of Iowa.