Abstracts

Rationale: Electrocorticography has proved to be a valuable technique for localization of epileptic foci. Many researchers have attempted to improve the sensitivity of ECoG recordings by redesigning electrodes with tighter electrode spacing and reduced electrode size in expectation of improved resolution. However, the depth of the neural sources is often neglected, leading to unintended consequences of these intuitive changes. This study focuses on estimating the trade-offs implicit in electrode size in order to provide a design with improved resolution at useful signal-to-noise ratios.Methods: An analytic model of the volume conductor was created and solved using electromagnetic reciprocity for the sensitivity of a sub-dural disc electrode to each point in the cortical volume. Sensitivity was measured as the recorded voltage due to a given source, relative to the recorded voltage of the same source at a different position, for example in Layer 5 relative to Layer 1, or at a depth of 1mm relative to directly on the cortical surface. ECoG recordings from 10 human subjects undergoing monitoring for intractable epilepsy were used to identify realistic electrode impedance, and signal-to-noise levels.Results: : Electrode disc diameter was found to affect not only the effective resolution of the recording, but also the depth of sensitivity in cortex. Thus, small electrodes are unable to record from all but the shallowest synapses of Layer I, while very large electrodes record activity averaged from Layer 1 to well past the neocortex. In order to maintain sensitivity of at least 50% through to Layer 5, an electrode with a diameter of at least 3.6mm was required. At this diameter, the sensitivity of an electrode in the plane parallel to the cortical surface fell to 50% at 3mm to 5mm depending on depth. If electrodes were spaced at this value, all points under the grid at the given depth would have at least 50% sensitivity to the nearest electrode relative to points with the highest sensitivity at that depth. This suggests traditional clinical grids (10mm spacing, 3mm electrode diameter) under-sample the cortical surface by ~2-3 times. In contrast, an electrode with diameter of 1mm requires an electrode spacing of only 1-4mm to maintain 50% sensitivity in the plane parallel to the cortical surface, but has only 26% sensitivity to layer 5 compared to layer 1.Conclusions: This model suggests theoretical and practical limits to the resolution of ECoG recordings (both over the cortical surface and in depth), and important parameters to consider to achieve useful recordings in explicit source volumes. Electrodes smaller than traditional clinical grids may not provide sensitivity to all layers of cortex, however the spacing of clinical grids can likely be improved, preventing areas between electrodes from being missed during mapping and localization. Precise identification of the volume of sensitivity of an electrode may allow researchers and clinicians to better analyze ECoG signals and to provide an improved understanding of the underlying neurobiology and pathology of these complex waves.