A Neural Network Based Approach to Whitening Biological Noise For Somatosensory Evoked Potential Detection

摘要

A matched filter is the optimal technique for detecting a known signal buried in additive white noise. Surface recorded somatosensory evoked potentials are corrupted with non-white noise, and so must be whitened before being passed to a matched filter. The authors evaluated the viability of selected artificial neural networks in whitening the noise of recorded evoked potentials.