Estimation of visual evoked potentials using a signal subspace approach

Extraction of visual evoked potentials (VEPs) from the human brain is generally very difficult due to its poor signal-to-noise ratio (SNR) property. A signal subspace technique is presented to estimate VEPs hidden inside highly colored electro-encephalogram (EEG) noise. This method is borrowed and modified from signal subspace techniques originally used for enhancing speech corrupted by colored noise. The signal subspace is estimated by applying eigenvalue decomposition on the approximated signal covariance matrix. The signal subspace-based algorithm is able to satisfactorily extract the P100, P200 and P300 peak latencies from artificially generated noisy VEPs. The simulation results show that the estimator maintains an average success rate of 87 % with an average percentage error of less than 9 %, when subjected to SNR from 0 dB to −10 dB.