Spike train SIMilarity Space (SSIMS): a framework for single neuron and ensemble data analysis.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Data Analysis

Background:

• Analyzing large-scale neural ensemble activity is crucial for understanding brain function.
• Existing methods like raster-histograms and pairwise correlations have limitations in capturing complex activity patterns.

Purpose of the Study:

• To develop a novel method for evaluating the relative similarity of neural spiking patterns.
• To enable visualization and clustering of large-scale neural activity beyond traditional analyses.

Main Methods:

• Combines spike train distance metrics with dimensionality reduction techniques.
• Uses vectors of pair-wise distances to represent relationships between neural activity patterns.
• Applies dimensionality reduction for concise data representation, clustering, and visualization.

Main Results:

• The spike train SIMilarity space (SSIMS) analysis successfully captured relationships between goal directions in a reaching task.
• SSIMS analysis segregated grasp types in a 3D grasping task without kinematic information.
• Demonstrated algorithm robustness and performance using multielectrode ensemble activity data from behaving primates.

Conclusions:

• The developed algorithm provides a powerful tool for exploring neural spiking data.
• Enables similarity-based clustering of neural activity states with minimal assumptions.
• Facilitates a deeper understanding of neural coding and brain circuit dynamics.