CUSP: Complex Spike Sorting from Multi-electrode Array Recordings with U-net Sequence-to-Sequence Prediction.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Machine Learning

Background:

• Complex spikes (CSs) in cerebellar Purkinje cells are crucial neural signals but are challenging to detect due to their variability and infrequency.
• Automated detection of CSs is hindered by waveform variability, low spike counts, and recording artifacts like electrode drift.

Purpose of the Study:

• To introduce CUSP (CS sorting via U-net Sequence Prediction), a novel deep learning framework for automated complex spike sorting.
• To enable accurate and robust detection and analysis of complex spike activity in high-density neural recordings.

Main Methods:

• CUSP utilizes a U-Net architecture with hybrid self-attention inception blocks for sequence-to-sequence prediction of CS events.
• The framework integrates local field potential and action potential signals for enhanced detection accuracy.
• Detected CS events are clustered and paired with simple spikes (SSs) to reconstruct Purkinje cell activity.

Main Results:

• CUSP achieves human-expert performance (F1 = 0.83 ± 0.03) in detecting complex spikes from macaque cerebellar recordings.
• The deep learning framework successfully identifies valid CS events missed during manual annotation.
• CUSP demonstrates superior performance compared to traditional and state-of-the-art algorithms, outperforming them in CS detection.

Conclusions:

• CUSP offers a scalable and robust solution for analyzing complex spike patterns in large-scale cerebellar and other neural datasets.
• The framework's robustness to waveform variability and electrode drift enables accurate long-term tracking of neural activity.
• CUSP provides a broadly applicable tool for studying neural information coding by combining expert-level accuracy with automation.