Real-time classification of experience-related ensemble spiking patterns for closed-loop applications

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Neural communication relies on temporally organized population activity patterns.
• Closed-loop manipulations are crucial for understanding neural patterns' causal roles in cognition.
• Existing brain-computer interfaces lack millisecond-timescale multi-neuronal pattern interpretation capabilities.

Purpose of the Study:

• To develop a system for classifying ensemble neural activity patterns in a closed-loop setting.
• To enable real-time identification of experience-related neural patterns.
• To bridge the gap in millisecond-timescale neural decoding for cognitive research.

Main Methods:

• Developed a system for classifying multi-neuronal spiking patterns at 10 ms resolution.
• Implemented a closed-loop system for online identification of neural activity.
• Demonstrated application in identifying hippocampal neuronal replay sequences in rats.

Main Results:

• Achieved online identification of experience-related patterns within 50 ms.
• Attained over 70% sensitivity and specificity in pattern identification.
• Classified the content of neural patterns with 95% accuracy.
• System demonstrated scalability to high-count electrode arrays.

Conclusions:

• The developed system effectively decodes multi-neuronal patterns at high temporal resolution.
• This technology facilitates the study of internally generated neural activity and its role in cognition.
• Advances in brain-computer interfaces are crucial for uncovering neural mechanisms underlying cognition.