A Comparative Study of Synchrony Measures for the Early Detection of Alzheimer’s Disease Based on EEG

Abstract

It has repeatedly been reported in the medical literature that the EEG signals of Alzheimer’s disease (AD) patients are less synchronous than in age-matched control patients. This phenomenon, however, does at present not allow to reliably predict AD at an early stage, so-called mild cognitive impairment (MCI), due to the large variability among patients. In recent years, many novel techniques to quantify EEG synchrony have been developed; some of them are believed to be more sensitive to abnormalities in EEG synchrony than traditional measures such as the cross-correlation coefficient. In this paper, a wide variety of synchrony measures is investigated in the context of AD detection, including the cross-correlation coefficient, the mean-square and phase coherence function, Granger causality, the recently proposed corr-entropy coefficient and two novel extensions, phase synchrony indices derived from the Hilbert transform and time-frequency maps, information-theoretic divergence measures in time domain and time-frequency domain, state space based measures (in particular, non-linear interdependence measures and the S-estimator), and at last, the recently proposed stochastic-event synchrony measures. For the data set at hand, only two synchrony measures are able to convincingly distinguish MCI patients from age-matched control patients (p < 0.005), i.e., Granger causality (in particular, full-frequency directed transfer function) and stochastic event synchrony (in particular, the fraction of non-coincident activity). Combining those two measures with additional features may eventually yield a reliable diagnostic tool for MCI and AD.KeywordsMild Cognitive ImpairmentGranger CausalityPhase SynchronyBlind Source SeparationMild Cognitive Impairment PatientThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.