Learning Stimulus Artifact Generation in Surface Recordings of Somatosensory Evoked Potentials
Surface recordings of somatosensory evoked potentials pose a challenging problem as the desired signal is obscured by stimulus artifact. A widely used approach for artifact reduction is adaptive noise cancellation, where an adaptive filter is used to map a primary signal to a reference signal. A major drawback of this technique is the dependency on temporal generalization. We propose a novel approach to artifact reduction that attempts to learn the process of artifact generation as the stimulus pulse amplitude increases.