When Wiener filtering is less than optimal: a illustrative application to the brain stem evoked potential

Wiener filtering as a means of improving the estimation of the wave form of the brain stem evoked potential (BSEP) is evaluated using both simulation and real signals. Where signal and noise overlap in the frequency domain, the performance of the filter is shown to be superior to signal averaging only at signal levels where the conventional approach already yields a working signal-to-noise ratio. It is concluded that, particularly given the computational overheads of Wiener filtering, the utility of the method must be doubted.