Automatic spike sorting for high-density microelectrode arrays.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Bioengineering

Background:

• High-density microelectrode arrays enable simultaneous recording of neural activity from thousands of neurons.
• Existing spike sorting algorithms struggle with the high dimensionality of data from these arrays.
• Efficient spike sorting is crucial for analyzing large neural datasets.

Purpose of the Study:

• To develop a scalable spike sorting algorithm for high-density microelectrode arrays.
• To overcome the "curse of dimensionality" in neural spike sorting.
• To create an automatic spike sorting method that requires no parameter adjustment or manual curation.

Main Methods:

• Sorting local electrode groups independently using classical spike sorting approaches.
• Employing data prewhitening before principal component analysis-based feature extraction.
• Utilizing a parameter-free clustering algorithm for spike classification.

Main Results:

• The algorithm effectively handles large data volumes from high-density arrays.
• Performance is competitive with state-of-the-art spike sorters in sensitivity and precision.
• The method is scalable to any number of recording electrodes and suitable for parallel computing.

Conclusions:

• The presented automatic spike sorting algorithm successfully scales classical techniques for high-density neural recordings.
• The combination of electrode grouping, subset clustering, and waveform prewhitening provides an efficient and accurate solution.
• This approach obviates the need for parameter tuning and manual intervention, making it a robust tool for neuroscientists.