Objective evaluation of differences in patterns of EEG topographical maps by Mahalanobis distance.

The purpose of this work was to develop a method to objectively evaluate differences in topographical patterns between 2 groups of isopotential maps. Previous evaluations by other investigators had been based on subjective judgment which created problems in describing differences in patterns, particularly when the differences were subtle. In the method presented in this report, the differences are evaluated by the Mahalanobis distance calculated from polynomial coefficient vectors obtained from unbiased polynomial interpolation in order to calculate values of picture elements in the 2-dimensional maps. One polynomial coefficient vector corresponds to one pattern of the map, containing both EEG and locational information. Examples of the evaluation are shown in the maps of fast activities induced by the injection of thiopental, diazepam or both drugs to an adult volunteer. In 5 experimental situations (A, control; B, thiopental; C, thiopental and diazepam; D, diazepam; and E, diazepam and thiopental administration), the polynomial coefficient vectors were obtained. When Mahalanobis distances (Q value) between the 2 maps for the 5 situations, arranged in 8 different combinations, were calculated, the values corresponded significantly to the pattern differences found between the maps. The greater the Q value, the more dominant the pattern difference. The objective evaluation method presented in this report has significant advantages over the subjective method previously reported and should be of interest to investigators attempting to quantify topographical pattern differences.