A subspace approach with pre-whitening for measurements of latencies in visual evoked potentials

Estimating a visual evoked potential (VEP) from the human brain is challenging since its signal-to-noise ratio (SNR) is generally very low—as low as −10 dB. Visual evoked potentials are conventionally extracted from the spontaneous brain activity by collecting a series of time-locked electroencephalogram (EEG) epochs and performing multitrial ensemble averaging on these samples to improve the SNR. Alternatively, a VEP estimation scheme based on a single VEP trial can be developed to reduce VEP recording time and minimize fatigue on subjects. In this paper, a subspace approach with pre-whitening (SAPW) has been proposed to estimate a single-trial VEP's P100 latency from colored EEG noise. The proposed method decomposes and decorrelates the corrupted VEP space into signal and noise subspace; VEP enhancement is achieved by removing the noise subspace and estimating the clean VEP only from the signal subspace. Since EEG is colored noise, explicit pre-whitening of the corrupted VEP waveform is performed in the proposed algorithm, to resolve diagonalization problems and achieve full VEP space decorrelation. The performance of SAPW in estimating VEP latencies has been assessed using simulated data at SNR ranging from 0 to −11 dB, and real patient data gathered in a hospital. The proposed SAPW estimator produces reasonably low failure rates and low average errors in both experiments.