Accurate identification of evoked potentials by waveform decomposition using discrete cosine transform modeling

This paper introduces a method for decomposing the component responses of the evoked potentials. The decomposition was realized by zero-pole modeling of the evoked potentials in the discrete cosine transform (DCT) domain. It was found that the DCT coefficients of a component response in the evoked potentials could be modeled sufficiently by a second order transfer function in the DCT domain. The decomposition of the component responses was approached by using partial expansion of the estimated model for the evoked potentials, and the effectiveness of the decomposition method was evaluated both qualitatively and quantitatively. Because of the overlap between the component responses, the proposed method enables an accurate identification of the component responses in the evoked potentials, which is useful for clinical and neurophysiological investigations.