A multi-region flexible neural interface for behavioral state decoding in freely moving mice.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Bioengineering

Background:

• Decoding whole-brain function requires high-density, long-term stable neural interfaces.
• Existing methods face challenges in scalability, long-term stability, and algorithm generalizability across time and subjects.

Purpose of the Study:

• To develop an integrated platform for accurate, stable, and generalizable decoding of behavioral states from whole-brain activity.
• To address limitations in current neural recording and decoding technologies.

Main Methods:

• Development of multi-region flexible probes (MRFPs) for distributed, long-term recordings (months) from 128 sites across eight brain regions.
• Implementation of a Conformer-based deep learning framework optimized for brain-wide neural dynamics.
• Comparative analysis of distributed versus concentrated electrode configurations for decoding performance.

Main Results:

• Achieved accurate (up to 89%) and stable decoding of behavioral states (resting, roaming, feeding, flash).
• Demonstrated that distributed neural sampling from five or more regions significantly improves decoding performance compared to concentrated configurations.
• Showcased robust generalization of decoding models across days and individuals without retraining.

Conclusions:

• The integrated platform provides a foundation for stable, high-fidelity multi-region electrophysiology.
• Offers a generalizable approach for decoding internal states from complex neural dynamics, advancing longitudinal and large-scale neuroscience studies.
• Enables progress in basic neuroscience research and the development of neural disorder therapies.