Computerized Topo-EEG Spectral Maps: Difficulties and Perspectives

Abstract

An electric potential measurement tells us only about the difference between two electrode locations. When prior experiments or present measurements prove that an electrode’s location is inactive (especially in ‘monopolar’ or scalp-to-reference technique), then EEG records and EEG maps can be interpreted regionally near each electrode location. Topographical mapping requires spatial interpolation which is one factor involved in spatial resolution. The interpolation algorithm has been studied by changing the power exponent n of the interelectrode distance d. When n is close to zero, interpolated values are all equal to the average of the four electric potential measurement Vs bounding the quadrilateral in which the interpolated points of the EEG map are located. When n is very high (n = 20 or higher), the values are essentially equal to the V which is at the minimum distance d. For n = 1, the map seems unlikely to represent the true scalp field. The choice between n = 2 and n = 3 is difficult, but n = 3 EEG mapping looks better (more plausible) and is our regular choice. The choice of recording/reporting method has also been studied. If the quantitative EEG map is linearly calculated from observations, the interpretation of record activity resulting from charge separation near the non-reference electrode attached to a particular channel is only possible when inactivity can be assured. This is completely different when the quantity mapped is not linearly calculated from observations (like power values or rms amplitudes in microvolts). As a result, they change their topographical shapes if the reference convention is changed, since a particular electrode used as a reference makes zero the power (or rms amplitude) around it. Following Lehmann, the least bad choice (but not the best, which is still unknown) seems to be the common average reference (instead of using the monopolar technique). We have computed EEG maps with different computerized references. Mapping interpretation should be made very carefully and cautiously. Reading and interpreting of scales are essential. Relative percent EEG maps present a percent non-dimensional scaling and would not change with a change of gain setting. More theoretical work is needed before reaching a normative consensus on EEG mapping rules.