Wavelets in neuroscience (Q2263179)

The complexity of brain processes makes it impossible to decipher neural signals using only methods of statistical analysis. The reaction of an organism to its environment is accompanied by changes in its dynamics, in the broader sense. Thus, neurophysiological signals should be regarded as time series of nonstationary processes produced by dynamical systems with unlimited degrees of freedom. It is this nonstationarity that makes wavelets a powerful tool for the analysis of recordings of neural dynamics, as they combine both localisation in time and a means of quantifying the frequencies present in a signal. This book gives an introduction to wavelets and then proceeds to demonstrate their use in a neural context. The first chapter gives a brief overview of some of the issues regarding signal processing in neuroscience, and the second gives an introduction to wavelets at a fairly low level. Chapter three concerns the use of wavelets to analyse recordings of the voltage within an individual cell (relative to the exterior), while Chapter four covers the use of wavelets in spike sorting: trying to partition a voltage signal recorded in the extracellular space into ``spikes'', each associated with one of a small number of nearby neurons. Chapter five moves on to the use of wavelets to study electroencephalogram (EEG) recordings and chapter six demonstrates this with applications to epileptic seizures and sleep spindles. Chapter seven is concerned with the automatic detection and classification of events within an EEG recording. Chapters three to seven all include many examples of applications to real data. The use of wavelets in the analysis of neural recordings provides significant benefits and technological advantages, but (as with any new technology) will take some time to be incorporated as standard. This book provides an excellent introduction to the area and should be of interest to anyone involved in analysing real neural recordings.