Our DeepFaceMouse paper is out
- 3 days ago
- 2 min read
Excited to share that our latest paper, "DeepFaceMouse enables scalable prediction of large-scale brain activity from facial dynamics in mice," has been published in Cell Reports Methods by Cell Press
This work started with a simple but fundamental question:
Can facial dynamics serve as a noninvasive window into brain-wide neural activity?
To address this challenge, we developed DeepFaceMouse, a deep learning framework that accurately tracks mouse facial movements and predicts large-scale cortical activity from behavior alone.
Key innovations of this study:
🔹 High-precision tracking of eyelid, whisker, nose, and mouth movements using deep learning
🔹 Prediction of large-scale cortical activity from facial dynamics, enabling scalable brain–behavior analyses
🔹 Validation across multiple calcium indicators, including GCaMP6s, GCaMP6f, and jGCaMP8m, demonstrating robust performance across diverse neural imaging conditions
🔹 Generalization across multiple behavioral paradigms, including spontaneous behavior, virtual reality navigation, and social interaction assays
🔹 Successful application across distinct neurological disease models, including glioblastoma (GBM), autism spectrum disorder (ASD), and multiple sclerosis (MS)
🔹 Open-source software and user-friendly GUI designed for large-scale neuroscience studies
Our findings suggest that subtle facial movements contain rich information about internal brain states and that artificial intelligence can help bridge the gap between observable behavior and large-scale neural dynamics.
More broadly, DeepFaceMouse reflects our lab's vision of integrating neuroscience, engineering, and AI to develop scalable tools for understanding brain function in health and disease.
A huge thank you to first author Kemal Ozdemirli and all of my trainees, collaborators, and co-authors who contributed to this work. Also big thanks to Cleveland Clinic Research, American Brain Tumor Association Case Comprehensive Cancer Center The National Institutes of Health for their support!
Comments