Virtual Pathologist

Image identification by machine learning models is a major application of artificial intelligence (AI). And, with ever-improving capabilities, the use of these models for medical diagnostics and research is becoming more commonplace. Doctors analysing X-rays and mammograms, for instance, are already being assisted by AI technology, and models trained to identify signs of disease in tissue sections are also being developed to help histopathologists. The models are trained with microscope images annotated by humans – the image, for example, shows a section of rat testis with signs of tubule atrophy (pale blue shapes) with other coloured shapes indicating normal tubules and structures. Once trained, the models are tasked with categorising unannotated datasets. The latest iteration of this technology was able to identify disease in testis, ovary, prostate and kidney samples with exceptional speed and high accuracy – in some cases finding signs of disease that even trained human pathologists had missed.

Written by Ruth Williams

Ruth Williams

Writer

Ruth is a freelance science journalist based in the US. She's a regular correspondent for The Scientist, and her work has appeared in The Lancet, Nature, Scientific American Mind, BBC Focus and elsewhere. Twitter @rooph

Website

• Image from work by Colin Greeley and colleagues
• Center for Reproductive Biology, School of Biological Sciences and the School of Electrical Engineering and Computer Science Washington State University, Pullman, WA, USA
• Image originally published with a Creative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0)
• Published in Scientific Reports, November 2024