The measurement of steady-state visual evoked cortical potentials using an adaptive noise canceller

Abstract

A new software method for measuring steady-state visual evoked cortical potentials (VECPs) has been developed using the principle of adaptive noise cancelling. The steady-state VECP is composed of narrow-band frequency components at harmonics of the stimulus frequency. In clinical recordings, these signal components are masked by wide-band noise, predominantly electroencephalographic activity and muscle noise. The stimulus frequency is exactly known and by using a reference sinusoid at the stimulus frequency (or its harmonics) the adaptive noise canceller (ANC) is able to cancel uncorrelated noise components from the recording. In effect, the ANC functions as an adaptive narrow-band-pass filter at the reference frequency. The performance of the ANC has been evaluated using both simulated and physiological signals. The output of the ANC provides temporal information on the signal amplitude and phase, and can be used to calculate the reliability of signal detection. For this application, the ANC has a number of advantages over the fast Fourier transform: it is a more sensitive detector, it requires fewer calculations, it is less computationally intensive, it requires less memory and it can be implemented in real time.