Single-Trial Estimation of Visual Evoked Potential Combining Wavelet Shrinkage and Adaptive Neural Network Filter

摘要

The Visual Evoked Potential (VEP) is a transient signal corrupted by the ongoing EEG noise, with poor SNR often below 0dB. To track the variation in the amplitude as well as latency of VEP signal, a modified Adaptive Neural Network Filter (ANNF) is employed in the estimation. In our frame work, the wavelet shrinkage technique is also used to improve the performance of the rejection of the ongoing EEG. Some simulation results illumine the possibility of single trial estimation.