Quantitative analysis of spatial sampling error in the infant and adult electroencephalogram.

The purpose of this report was to determine the required number of electrodes to record the infant and adult electroencephalogram (EEG) with a specified amount of spatial sampling error. We first developed mathematical theory that governs the spatial sampling of EEG data distributed on a spherical approximation to the scalp. We then used a concentric sphere model of current flow in the head to simulate realistic EEG data. Quantitative spatial sampling error was calculated for the simulated EEG, with additive measurement noise, for 64, 128, and 256 electrodes equally spaced over the surface of the sphere corresponding to the coverage of the human scalp by commercially available "geodesic" electrode arrays. We found the sampling error for the infant to be larger than that for the adult. For example, a sampling error of less than 10% for the adult was obtained with a 64-electrode array but a 256-electrode array was needed for the infant to achieve the same level of error. With the addition of measurement noise, with power 10 times less than that of the EEG, the sampling error increased to 25% for both the infant and adult, for these numbers of electrodes. These results show that accurate measurement of the spatial properties of the infant EEG requires more electrodes than for the adult.