radRounds Radiology Network

Connecting Radiology | Enabling collaboration and professional development

Can a Pigeon Read a Mammogram Better than a Radiologist?

Not only are pigeons good at getting stuck in air shafts, they also have phenomenal vision. A group of researchers with diverse backgrounds from the University of California, Davis, Emory University, and the University of Iowa, have found that pigeons are skilled at distinguishing benign from malignant breast cancer tumors in mammograms.

In a dark room, the researchers trained the birds to peck at a series of images and choose their corresponding category buttons. The pigeons made several “observing pecks”, and afterwards were prompted to peck at either a yellow or blue button associated with the category of image. If they pecked correctly, they were awarded with food.

It took more than just a few days for the pigeons to learn how to accurately categorize the different images. Their rate at correctly distinguishing between benign and malignant breast cancer histopathy increased from 50 percent on day one to 85 percent by days 13 through 15. In order to determine that birds were not just memorizing the images, the researchers presented them a mix of old and new slides at varying levels of magnification.

Although the pigeons correctly assessed the vast majority of images, they struggled to correctly diagnose certain benign images. For example, slides that featured benign breast lobular structures were “highly cellular and densely packed” and often confused to be malignant. The slides that contained “duct-like structures” resembled common benign details. The researchers write that their errors are not arbitrary, the images they misinterpreted were simply very different from the rest of the set.  

“The birds’ successes and difficulties suggest that pigeons are well-suited to help us better understand human medical image perception, and may also prove useful in performance assessment and development of medical imaging hardware, image processing, and image analysis tools,” write the researchers.

Views: 734


You need to be a member of radRounds Radiology Network to add comments!

Join radRounds Radiology Network

Sponsor Ad

© 2020   Created by radRounds Radiology Network.   Powered by

Badges  |  Report an Issue  |  Terms of Service