Spatial resolution of dipole sources in EEG recordings

Abstract

The starting point of this thesis is the observation of Freeman and Baird (1987) that in rabbits, it is possible to record from the surface of the brain spatial patterns of voltage that successfully classify individual animals as either experiencing or not experiencing various olfactory, auditory and visual sensations. The first question asked in the thesis is “What sort of pattern of neural activity would give rise to electrocorticogram patterns with the spatial frequency measured in rabbits?” On the assumption that electric sources in the cortex can be modeled as equivalent dipoles, both mathematical modeling and experimental measurements of the fields produced by dipoles in saline are used to show that the relevant patterns would be produced by dipoles located approximately 1 – 2 mm apart in the cortex. The next question is, “If such voltage patterns were to occur in humans, would it be possible to resolve them in EEG recordings taken from the scalp?” The answer is that, even ignoring the smearing effects of the skull and using a high pass spatial filter to aid resolution (it is demonstrated that a Laplacian filter significantly increases the source-electrode distance at which pairs of dipoles can be resolved), it is not possible to resolve two dipoles positioned 2 mm apart if the recording array is more than 2.5 mm above the dipole pair. Since the average distance between the surface of the brain and the scalp in humans is 15 mm (McConnell et al 2001), this strongly suggests that it will never be possible to resolve patterns like those recorded by Freeman and Baird (1987) from scalp EEG records in humans. Even using a high pass spatial filter and ignoring the effects of the skull, two dipoles have to be more than 15 mm apart before they can be resolved from 15mm away. The conclusion is that, if Freeman and Baird’s patterns do represent neural correlates of consciousness (NCCs), it will not be possible to record human NCCs from the scalp. Dural or sub-dural recording will be necessary.