IMPROVED REALISTIC LAPLACIAN ESTIMATE OF HIGHLY-SAMPLED EEG POTENTIALS BY REGULARIZATION TECHNIQUES

Ln this study vile investigated the effects of lambda correction, gene ralized cross-validation (GCV), and Tikhonov regularization techniques on the realistic Laplacian (RL) estimate of highly-sampled (128 chann els) simulated and actual EEG potential distributions. The simulated E EG potential distributions were mathematically generated over a 3-shel l spherical head model (analytic potential distributions). Noise was a dded to the analytic potential distributions to mimic EEG noise. The m agnitude of the noise was 20, 40 and 80% that of the analytic potentia l distributions. Performance of the regularization techniques was eval uated by computing the root mean square error (RMSE) between regulariz ed RL estimates and analytic surface Laplacian solutions. The actual E EG data were human movement-related and short-latency somatosensory-ev oked potentials. The RL of these potentials was estimated over a reali stically-shaped, magnetic resonance-constructed model of the subject's scalp surface. The RL estimate of the simulated potential distributio ns was improved with all the regularization techniques. However, the l ambda correction and Tikhonov regularization techniques provided more precise Laplacian solutions than the GCV computation (P < 0.05); they also improved better than the GCV computation the spatial derail of th e movement-related and short-latency somatosensory-evoked potential di stributions. For both simulated and actual EEG potential distributions the Tikhonov and lambda correction techniques provided nearly equal L aplacian solutions, but the former offered the advantage that no preli minary simulation was required to regularize the RL estimate of the ac tual EEG data. (C) 1998 Elsevier Science Ireland Ltd.