Comparison of methods for extracting of evoked potentials

In this work the algorithms are divided into three groups. First group consists of one method as Subspace Regularization Method (SRM). Second group consists of the combination both Wiener Filter (WF) and Subspace Method (SM) and Coherence Weighted Wiener Filter (CWWF) and SM in different order. The application of SM to filtered data with WF is proposed in literature, in this study we propose the reverse of this method using with the CWWF instead of WF. Third group consists of adaptive filters (Adaptive Kalman Filter (AKF), RLS and LMS). We propose two new methods including the combination of SM and AKF and SM and LMS Filter in 3rd group. All methods belong to 2nd and 3rd group are compared to each other and then best of each group are compared to SRM and Ensemble Averaging (EA) quantitatively. The highest SNR values are supplied with SRM among all methods mentioned in this work for small number of trials. However, the combination both SM and CWWF and the combination both SM and AKF provide higher output SNR values than EA, also. All the combination both SM and one method increase the advantages than each of them separately.