New Tool Tracks Mouse Facial Expressions

Facial expressions convey a full range of emotions—love, pain, joy, fear—that we intuitively understand. However, precisely measuring the complex relationship between these expressions and brain activity has remained elusive, especially in mice with their distinct cone-shaped faces compared to humans. Helen Hou and colleagues from Cold Spring Harbor Laboratory (CSHL) introduced Cheese3D, a new discovery platform published in Nature Neuroscience, to bridge this gap.

Cheese3D uses a sophisticated camera and computer vision system to detect even subtle changes in mouse facial movements. AI processes these observations into quantifiable data, allowing researchers to systematically analyze brain-behavior links. The idea stemmed from necessity, as Hou recalls: “When I started my lab, we were really excited to capture the rich repertoire of facial behavior.” While veterinarians can assess animal well-being from faces, no prior automated tool provided the resolution needed for brain insights.

CSHL has long advanced mice as models for neuroscience, but their facial anatomy posed challenges. Partnering with CSHL’s Core Facilities, the Hou lab created a system of six tiny cameras capturing simultaneous multi-angle footage. Machine learning assembles the videos expertly while monitoring brain electrical activity.

Validation involved real behaviors like eating and, notably, anesthesia. Cheese3D gauged wakefulness depth with precision matching gold-standard EEG, achieved non-invasively in collaboration with CSHL’s Borniger lab. Hou emphasizes: “Very subtle changes in facial muscle tone teach us a lot. So, we can predict depth of anesthesia in a non-invasive way using the face.”

This new tool extends to clinical potential. Hou is examining facial expressions in disease states and developmental milestones, like smiling before crawling or walking. Such work could advance autism and behavioral therapy understanding.