Recent studies have utilized pigeons to assess their ability to recognize medical images, displaying varying degrees of success. In a study published in 2015, pigeons trained to identify breast tissue images achieved a remarkable pooled accuracy of 99% in categorizing benign and malignant slides. However, they struggled with new images and could only memorize training examples, failing to classify unfamiliar breast masses effectively. Fast forward to 2026, two new studies revealed mixed results. One study found that pigeons could classify echocardiogram videos using an AI overlay that highlighted critical areas but could not perform well when the overlay was removed. In contrast, another study showed that pigeons successfully learned to detect solid lung nodules in CT scans and transferred this skill to other abnormalities. These studies suggest pigeons can mimic some human strengths and weaknesses in image reading, but the researchers emphasized that the findings do not imply pigeons can assist in clinical diagnostics. The work illustrates the potential use of pigeons for affordable research and insights into machine learning but should not change current medical screening practices.
Mon, 26 Oct 2026 12:00:18 GMT | Medical Daily