A Five-Channel Weighted Real-Time Algorithm for High-Density Electrodes Spike Sorting.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• High-density neural probes enable multi-channel neuron detection, surpassing traditional single-channel limitations.
• Accurate spike sorting is crucial for analyzing neural signals from dense probe recordings.

Purpose of the Study:

• To develop and evaluate a real-time, five-channel weighted spike sorting algorithm for high-density neural probes.
• To improve the accuracy of spike sorting by incorporating information from adjacent neural channels.

Main Methods:

• A five-channel template is created by weighting signals from a center channel and four adjacent channels.
• A modified OSort algorithm with unsupervised learning is used for real-time template updating.
• The algorithm was tested using ground truth recordings and simulated neural datasets.

Main Results:

• The proposed five-channel algorithm achieved an average sorting accuracy of 89%.
• This represents a significant improvement over traditional single-channel spike sorting, which achieved 78% accuracy.
• Utilizing adjacent channel information demonstrably enhances sorting precision.

Conclusions:

• The five-channel weighted spike sorting algorithm offers superior performance for high-density neural probe data.
• This method provides a more accurate and reliable approach to analyzing neural activity in complex recording environments.
• The real-time, unsupervised learning capability makes it suitable for online spike sorting applications.