Key Takeaways
• Cheese3D uses AI and multi camera imaging to track detailed 3D facial movements in mice
• The system captures facial activity with sub millimeter precision at high speed
• It can predict brain states like depth of anesthesia without invasive procedures
• Matches the accuracy of traditional EEG methods while keeping animals undisturbed
• Could open new research paths in pain assessment, brain disorders, and behavior
Researchers at Cold Spring Harbor Laboratory have introduced a groundbreaking system that uses artificial intelligence to read facial expressions in mice and translate them into insights about brain activity. The technology, called Cheese3D, offers a powerful new way to study the brain without relying on invasive techniques.
The study, published in Nature Neuroscience, was led by Helen Hou, an assistant professor at the laboratory. Her team set out to capture the subtle and often overlooked facial movements that reflect an animal’s internal state.
Cheese3D uses six miniature cameras positioned around a mouse to record its face from multiple angles at the same time. Advanced machine learning models then reconstruct a precise three d imensional representation of the animal’s facial movements. This includes fine details such as ear position, eye motion, whisker movement, and jaw activity on both sides of the face.
The system achieves sub millimeter spatial accuracy and operates at a temporal resolution of 100 frames per second. This level of detail allows researchers to detect extremely subtle changes in facial muscle tone that would otherwise go unnoticed.
What makes the system particularly valuable is its ability to link these facial movements directly to brain activity. While recording facial data, the researchers simultaneously monitored neural signals, enabling them to correlate external expressions with internal brain states.
In early experiments, the team used Cheese3D to study mice under anesthesia. By analyzing facial features, the system was able to accurately determine how deeply the animals were asleep or awake. The results closely matched those obtained using traditional electroencephalography, or EEG, methods, which typically require more intrusive setups.
According to Helen Hou, even the smallest shifts in facial muscle tone can reveal meaningful information about the brain. This opens the possibility of assessing neurological states in a completely non invasive way, reducing stress on research animals while improving data quality.
The implications extend far beyond anesthesia monitoring. Hou and her collaborators are now exploring how facial expressions change in disease conditions, including those related to pain and neurological disorders. The approach could also help researchers better understand how facial expressions develop and are learned over time.
Because facial movement is one of the earliest behaviors seen in human development, insights from this research could have broader applications. Understanding how social expressions emerge may contribute to studies on developmental conditions such as autism and inform future behavioral therapies.
Importantly, the Cheese3D platform has been released as an open source Python package, complete with visualization tools. This makes the technology accessible to laboratories worldwide, potentially accelerating discoveries in neuroscience and behavioral science.
As AI continues to bridge the gap between observable behavior and internal brain processes, systems like Cheese3D highlight a future where complex biological signals can be decoded through simple, non invasive observation.
Your First 10 AI Skills
10 practical AI skills, copy-paste prompts and a 7-day plan to start using AI with confidence.
Download the guide →