Dynamical Nonstationarity of EEGs in patients with AD/HD

Abstract

In the specific application domains of psychiatry and cognitive neuroscience the need of new diagnostic tests are crucial, because there are too few and not always effective, especially in the Attention-Deficit/Hyperactivity Disorder, a major mental disorder occurring during childhood. AD/HD patients are particularly hard to characterize, even though they present specific activity pattern of the brain, they present many subtypes and comorbidity, making them comparable to many others pathologies such as anxiety or depression disorder. Based on the assumption that the brain is a deterministic nonlinear dynamical system, we engaged a dynamical analysis of its EEG recordings. Such analyses are becoming more and more common, and use the recent theory of Chaos and Nonlinear dynamical systems analysis. Our major Hypothesis being the characteristic instability of attention of the patients under investigations (AD/HD), we tried to quantify this attention deficit using the EEGs Time Series Recordings from the scalp by operating a Dynamical Quasi-stationary segmentation of the Time Series. Our assumption also includes the involvement of a specific cognitive function regulating attention, the so called Executive function, and then we expect a major activation of the frontal regions. In this report, we have developed a nonlinear method to extract quasi-stationary segment from a time series, and we apply this method to EEG time series. The quasi-stationary time segmentation is used as a new feature to quantify the time series, or from a neuro-scientific point of view, to quantify some cognitive states, represented by the EEG signal. We studied two populations, control subjects and AD/HD patients. We have found, indeed that there are significant differences between AD/HD patients and control subjects, especially in the frontal and the temporal regions. Thus we demonstrated that the dynamical nonstationarity analysis is able to quantify the attention in our problem, and...