Suivi de relation en temps et en fréquence dans les signaux SEEG épileptiques. Evaluation de deux estimateurs linéaires

This article deals with the tracking of the evolution in time and frequency of the linear relationship between two nonstationary signals (intracerebral EEG signals recorded in epileptic patients). An estimator is proposed, based on the measurement of the linear correlation coefficient between sub-band signals, optimized for the time delay. This estimator is compared to an estimator of the coherence function (traditionally used in the analysis of EEG signals). Results on simulated signals show that with the novel estimator we can reduce the bias and the variance of estimation. In the case of real signals, the use of proposed estimator makes it possible to reveal a difference, important from the physiopathology point of view, between the correlation of signals corded during the interictal activity (between seizures) and during the ictal activity (during seizures). re