Noise reduction of chaotic interspike intervals (ISI) of neurons

Complex firing patterns of neurons have received great interest. It is known that neurons work in notoriously noisy environments, as a result, the ISI time series measured in experiments contain a high level of noise. Although research work shows that noise may play a positive role in detection of a weak signal for sensory neurons, it is indeed a nuisance for analysis of irregular ISI data using traditional chaotic time series methods which are rather sensitive to noise, for noise can mask deterministic dynamics underlying the ISI signals and the wrong conclusion will be drawn based on those tests. We first report that noise can be effectively removed from ISI time series both from a theoretical model and experimental recordings with just a simple nonlinear noise reduction scheme and deterministic dynamics underlying irregular ISI data will clearly be recovered