Pigeons Used To Spot Cancer On Mammograms

Scientists say they could help researchers explore quality and the impact of colour, contrast, brightness and image compression on diagnostic performance.

But they found it more hard to classify suspicious masses on the scans – a task described as “very challenging” even for expert humans.

Among the images of actual tissue samples, the birds’ accuracy rose from 50 percent (equivalent to chance) to about 85 percent 15 days later. Pigeons possess neural pathways strikingly similar to those at work in the human brain, said the scientists.

They proved especially adept at sorting out the slide samples, the scientists found.

In the third experiment – which entailed identifying masses in mammograms – the birds were put in the tricky role of the radiologist and asked to identify cancerous masses that have less definition than in the tissue slides. The pigeons were able to do identify masses in images they had seen before, but when they were shown new mammograms they could not identify masses as benign or malignant with any accuracy.

They trained the birds to tell the difference between healthy tissue and tumor tissue as displayed on the slides that pathologists use to look at biopsies in the lab. The birds peck a colored square in a computer screen to show when they’ve decided whether a sample is cancer or benign tissue.

Co-author Professor Edward Wasserman, from the University of Iowa, US, said: “These results go a long way toward establishing a profound link between humans and our animal kin”.

He said: “The pigeons learned to discriminate benign from cancerous slides as fast in this research as in any other study we’ve conducted on pigeons in our laboratory”.

In the mammogram study their accuracy averaged 84 per cent for images with microcalcifications they had been trained upon and 72 per cent for novel images – on a par with human radiologists who were given the same cases to review.

As with humans, the accuracy of the pigeons was affected by the presence or absence of colour in the images, and degrees of image compression. The birds could do that job instead, they suggest.

A few doctors with a vast amount of training and years of education can misinterpret mammogram scans and have a hard time interpreting microscopic slides.

“The remarkable thing is the pigeons did just as good on the new images – which means they had learned something, they had not just memorized what they had been shown”, Levenson said.

“Pigeons’ sensitivity to diagnostically salient features in medical images suggest that they can provide reliable feedback on many variables at play in the production, manipulation, and viewing of these diagnostically crucial tools”, added Levenson.

“This is a hard, time-consuming, and expensive process that requires the recruitment of clinicians as subjects for these relatively mundane tasks”, he said.

Additional researchers from Emory University and the University of Iowa contributed to the study, which appears in PLOS ONE.