A new neuron spike sorting method using maximal overlap discrete wavelet transform and rotated principal component analysis

摘要

A new method for neuron spike sorting is presented that uses maximal overlap discrete wavelet transform (MODWT) and rotated principal component analysis. MODWT is very effective in extracting the bandwidth of neuron firings without shape distortion. We then used a rotated principal component analysis to isolate unique neuron templates. In this procedure the first principal component serves as a neuron spike template. This component is then removed from the original data and the procedure is repeated. Thus, the recursive procedure relaxes the orthogonality constraints of traditional PCA. This method provides a reliable technique for discriminating neuron spike firings from the original signal itself.