Non-stationary analysis of extracellular neural activity

Activity in many neural signals is often non-stationary and has a broad Frequency spectrum. We present a discrete wavelet transform based method for determining the energy profile in such neural signals. The method can be used to examine changes in the energy profile of the signal in response to perturbations and for extraction of different rhythmic components of the neural signal. We test the efficacy of our method by using synthesized signals with additive white noise. Application of the method is illustrated by determining the energy profile and extracting the locomotor burst in neural signals generated by spinal motor networks of the neonatal rat. (C) 2000 Elsevier Science B.V. All rights reserved.