Automatic extracellular spike denoising using wavelet neighbor coefficients and level dependency

Extracellular spike denoising is a technical challenge because of large amounts of background noises and contributions of many neurons to recorded signals. In this paper, a wavelet denoising using neighbor coefficients and level dependency was proposed to separate spikes from background noise. This method is based on the inter- and intra-scale correlation of neighboring wavelet coefficients to select those that deviate from baseline noise. The performance of the method was evaluated in both simulated signals with different noise levels and real neural recordings from primary visual cortex of rat, and the results were compared with several previously proposed wavelet-based methods and the noisy spike signals. The results of simulated and real data show that the signal-to-noise ratio of spikes was considerably improved by the proposed method and the number of false positive was significantly reduced.