A team from ETH Zurich has presented GrimACE, an artificial intelligence system that analyzes the facial expressions and postures of mice to detect signs of pain or discomfort. Until now, this assessment depended on the human eye, with subjective and inconsistent results. The new tool promises to standardize animal welfare in research.
Computer vision to measure silent suffering 🐭
GrimACE uses computer vision and machine learning to examine subtle changes in the rodents' physiognomy: narrowing of the eyes, bulging of the nose and cheeks, or alterations in the ears and whiskers. It also analyzes body posture. The system processes videos in real time, assigning objective scores that replace subjective observation. Researchers thus aim to reduce animal stress and improve data reproducibility.
Mice with no secrets: the AI that reads your face (and ears) 🧠
Now mice won't be able to pretend they're fine when something hurts them. GrimACE exposes them with the same effectiveness as a friend who tells you that you look unwell. Rodents, which until now hid their discomfort with stoicism, face a system that is not fooled by a stiff tail or a drooping whisker. Science advances, and mice will have to try harder in their next performance.