Quantification of synchronization processes by coherence and phase and itsapplication in analysis of electrophysiological signals

Neuronal activity during information processing and muscle activity are gen erally characterized by oscillations. Mostly, widespread areas are involved and electrophysiological. signals are measured on different sites of the c ortex or of the muscle. In order to investigate functional relationships between different componen ts of multidimensional electrophysiological signals, coherence and phase an alyses turned out to be useful tools. These parameters allow the investigat ion of synchronization phenomena with regard to oscillations of defined fre quencies or frequency bands. Coherence and phase are closely connected spectral parameters. Coherence ma y be understood as a measure of phase stability. Whereas coherence describe s the amount of common information with regard to oscillations within certa in frequency bands, the corresponding phase, from which time delays of thes e oscillations can be computed, hints at the direction of information trans fer through oscillation. Coherence and phase analysis of surface EMG during continuous activity of d eep and superficial muscles show distinct differences due to volume conduct ion properties of myoelectrical signals. Superficial activity therefore is characterized by significant coherence and stable phase relationships, whic h, additionally, can be used to determine motor unit action potential (MUAP ) propagation velocity along the fibre direction without application of inv asive methods. Deep muscle activity lacks significant coherence. Mental processes can be very brief and cooperation between different areas may be highly dynamic. For this reason in addition to usual Fourier estimat ion of coherence and phase, a two-dimensional approach of adaptive filterin g was developed to estimate coherence and phase continuously in time. Stati stical and dynamic properties of instantaneous phase are discussed. In order to demonstrate the value of this method for studying higher cognit ive processes the method was applied to EEG recorded during word processing . During visual presentation of abstract nouns an information transfer thro ugh the propagation of oscillations from visual areas to frontal associatio n areas in the alpha (1)-frequency band could be verified within the first 400 ms. In contrast, in case of auditory presentation positive phases from the temporal electrode locations T3 and T4 towards the occipital areas appe ar within the time interval of 300 ms-600 ms. The alpha (1)-band predominat ely seems to reflect sensory processing and attention processes.