Reconstruction of intracortical whisker-evoked local field potential from electrocorticogram using a model trained for spontaneous activity in the rat barrel cortex

Abstract

Electrocorticogram (ECoG) has provided neural information from the cortical surfaces, is widely used in clinical applications, and expected to be useful for brain-machine interfaces. Recent studies have defined the relationship between neural activity in deep layers of the cerebral cortex and ECoG. However, it is still unclear whether this relationship is shared across different brain states. In this study, spontaneous activity and whisker-evoked responses in the barrel cortex of anesthetized rats were recorded with a 32-channel ECoG electrode array and 32-channel linear silicon probe electrodes, respectively. Spontaneous local field potentials (LFPs) at various depths could be reconstructed with high accuracy (R>0.9) by a linear weighted summation of spontaneous ECoG. Current source density analysis revealed that the reconstructed LFPs correctly represented laminar profiles of current sinks and sources as well as the raw LFP. Moreover, when we applied the spontaneous activity model to reconstruction of LFP from the whisker-related ECoG, high accuracy of reconstruction could be obtained (R>0.9). Our results suggest that the ECoG carried rich information about synaptic currents in the deep layers of the cortex, and the same reconstruction model can be applied to estimate both spontaneous activity and whisker-evoked responses.

Keywords: BMI; Current source density; Decoding; LFP; Micro-ECoG; Sparse linear regression.