Generalized Laminar Population Analysis (gLPA) for Interpretation of Multielectrode Data from Cortex

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Laminar Population Analysis (LPA) decomposes electrical data from linear multielectrodes in the cortex.
• LPA utilizes physiological constraints, linking Local Field Potentials (LFP) to Multi-Unit Activity (MUA), unlike purely mathematical methods like PCA or ICA.
• Existing LPA methods may be improved by extending their basis functions.

Purpose of the Study:

• To introduce and evaluate a generalized Laminar Population Analysis (gLPA) method.
• To assess the performance of gLPA in decomposing cortical electrical signals using biophysical models.
• To determine the optimal number of postsynaptic LFP kernels for gLPA.

Main Methods:

• Developed generalized Laminar Population Analysis (gLPA) with an extended set of basis functions.
• Generated synthetic Local Field Potential (LFP) data using biophysical forward-modeling of a thalamocortical network model (Traub model).
• Tested various gLPA versions (gLPA-2, gLPA-3) against known ground truth data to evaluate performance and mitigate overfitting.

Main Results:

• The original LPA method showed fair agreement with ground-truth laminar components.
• gLPA, particularly with two (gLPA-2) or three (gLPA-3) postsynaptic LFP kernels, demonstrated improved accuracy in extracting laminar components.
• gLPA provided a better fit to experimental data compared to the original LPA.

Conclusions:

• The gLPA method offers enhanced accuracy for analyzing laminar population activity in the cortex.
• Utilizing multiple postsynaptic LFP kernels in gLPA significantly improves the decomposition of electrical signals.
• gLPA provides a more refined approach to understanding cortical population dynamics from electrophysiological recordings.