Frontiers in Systems Neuroscience (Apr 2014)
Modeling and analysis of extracellular potentials in multielectrode arrays.
Abstract
Multielectrode array (MEA) measurements of neural activity in brain slices are becoming more and more ubiquitous. To properly interpret the results gathered with their help, it is important to understand the measurement physics - the link between the electrical potential recorded at the electrodes and the underlying neural activity. In this work, we modeled the generation of transmembrane currents in a slice preparation and the extracellular potentials they generate. We addressed this forward modeling problem using Finite Element Methods (FEM) and used these results as ground truth for comparison with simplified modeling schemes in particular, forward modeling by means of the analytical point- and line-source formulas amended by means of the “method of images” (MoI) . To compare the different schemes we tested them on current sources and tissue models of increasing complexity. We calculated the extracellular potential generated at the MEA plane while varying the following: 1) We placed a point current source inside the slice, and varied its position with respect to the MEA plane. We also varied the conductivity profile of the slice to include inhomogeneity, anisotropy, and a thin layer of saline beneath the slice. 2) We used a layer 5 pyramidal cell model (Hay et al., 2011) and tracked the transmembrane currents through its compartments during a spontaneous spike. We placed these transmembrane currents as line current sources on a reconstructed cell within the slice. We then varied the conductivity of the saline bath surrounding the slice. 3) For testing a population of cells, we used a thalamo-cortical column model (Traub et al., 2005), and tracked the transmembrane currents through all its cortical neurons during an evoked response. The MEA potentials were then calculated assuming point-like current sources positioned on the mid points of the compartments in the multicompartment neuron models. We varied the conductivity of the slice to include inhomogeneity, anisotropy, and a thin layer of saline beneath the slice . In all cases we found that the MoI gave quantitatively accurate results compared to FEM. We further investigate how the changes in the extracellular potential due to the variations in the conductivity profile of the slice affect the estimates of current source density (CSD) in a MEA setup. We used MoI to extend the kernel current source density method (Potworowski et al., 2012) to include the geometry of the slice and saline slab. We found that including the saline layer does not increase the accuracy of the estimated CSD.
Keywords
