Effect of Using Suboptimal Coefficients for Laplacian Estimation with T-Lead Electrodes on Human Electroencephalogram Data and Via Finite Element Method Modeling

Recent research showed that optimal coefficients (6, -1) maximizing the accuracy of Laplacian estimation via tripolar concentric ring electrode with dimensions similar to those of t-Lead are different from currently used coefficients (16, -1). To assess the impact of this difference, linear time and frequency domain (cross-correlation and coherence respectively) synchrony measures applied to resting electroencephalogram from six healthy subjects revealed high signal synchrony in tripolar Laplacian estimates indicating that the difference between optimal and suboptimal coefficients may not be significant. Aims of this study were to add finite element method (FEM) modeling and a nonlinear signal synchrony measure (normalized mutual information, NMI) to further assess the difference between the optimal and suboptimal coefficients. Together the mean NMI value of approximately 0.7 and FEM modeling results with mean normalized maximum error ratio of 3.57 and mean normalized spatial selectivity ratio of 1.54 suggest that the difference between using the optimal and suboptimal coefficients for Laplacian estimation is significant.