The blessing of Dimensionality: Feature Selection outperforms functional connectivity-based feature transformation to classify ADHD subjects from EEG patterns of phase synchronisation.

Functional connectivity (FC) characterizes brain activity from a multivariate set of N brain signals by means of an NxN matrix A, whose elements estimate the dependence within each possible pair of signals. Such matrix can be used as a feature vector for (un)supervised subject classification. Yet if N is large, A is highly dimensional. Little is known on the effect that different strategies to reduce its dimensionality may have on its classification ability. Here, we apply different machine learning algorithms to classify 33 children (age [6-14 years]) into two groups (healthy controls and Attention Deficit Hyperactivity Disorder patients) using EEG FC patterns obtained from two phase synchronisation indices. We found that the classification is highly successful (around 95%) if the whole matrix A is taken into account, and the relevant features are selected using machine learning methods. However, if FC algorithms are applied instead to transform A into a lower dimensionality matrix, the classification rate drops to less than 80%. We conclude that, for the purpose of pattern classification, the relevant features should be selected among the elements of A by using appropriate machine learning algorithms.

The variability of EEG functional connectivity of young ADHD subjects in different resting states.

OBJECTIVE: To assess ADHD from global measures of EEG functional connectivity and their temporal variability in different resting states. METHODS: EEGs from sixteen cortical regions were recorded at rest during eyes-closed (EC) and eyes-open (EO) in 10 male combined-type ADHD subjects and 12 healthy male controls. The mean global connectivity (CM) of each region and its temporal variability (CV) were estimated from a number of EEG segments recorded in both states. Connectivity indices between regions were calculated using the magnitude squared coherence (Coh) in the delta(δ)/theta(θ)/alpha(α)/beta(ß) frequency bands and the nonlinear index (L) of generalized synchronization. RESULTS: The CM did not present between-group differences in any region or state. However, the CV exhibited state-independent differences between both groups (ADHD>controls) mainly in frontal and parieto-occipital regions for all indices except Coh(α). Within group, only the CV-Coh(θ) of the centro-temporal region increased significantly for the ADHD subjects from EC to EO (p<0.001) and was greater than controls in EO (p<0.001). CONCLUSIONS: The CV of index-L and of Coh(θ) seem to be the best state-independent and -dependent measurements, respectively, to discriminate ADHDs from control subjects using resting state EEG data. SIGNIFICANCE: The underlying neural dysfunctions producing the ADHD seem better reflected by the CV measurements.

Electroencephalography signatures of attention-deficit/hyperactivity disorder: clinical utility.

The techniques and the most important results on the use of electroencephalography (EEG) to extract different measures are reviewed in this work, which can be clinically useful to study subjects with attention-deficit/hyperactivity disorder (ADHD). First, we discuss briefly and in simple terms the EEG analysis and processing techniques most used in the context of ADHD. We review techniques that both analyze individual EEG channels (univariate measures) and study the statistical interdependence between different EEG channels (multivariate measures), the so-called functional brain connectivity. Among the former ones, we review the classical indices of absolute and relative spectral power and estimations of the complexity of the channels, such as the approximate entropy and the Lempel-Ziv complexity. Among the latter ones, we focus on the magnitude square coherence and on different measures based on the concept of generalized synchronization and its estimation in the state space. Second, from a historical point of view, we present the most important results achieved with these techniques and their clinical utility (sensitivity, specificity, and accuracy) to diagnose ADHD. Finally, we propose future research lines based on these results.

Performance analysis of univariate and multivariate EEG measurements in the diagnosis of ADHD.

OBJECTIVE: To investigate the performance of univariate and multivariate EEG measurements in diagnosing ADHD subjects in a broad age range. METHODS: EEG from eight cortical regions were recorded at rest during eyes open and eyes closed in 22 male ADHD subjects of combined type and 21 healthy male controls (age range 4-15 years). Univariate and interdependence measurements calculated from the frequency domain and from the reconstructed state spaces of EEG signals were computed, and their performance in discriminating ADHD from healthy subjects was analyzed. RESULTS: Significant between-group differences in univariate measures were age-dependent. However, certain interdependence inter-hemispheric measures during eyes closed showed significant, age-independent between-groups differences. Among them, coherence in the beta band between inter-occipital regions and between left/occipital-right/central regions provided an overall accuracy classification rate of 74.4%. Even greater accuracy (86.7%) was obtained by an interdependence index of generalized synchronization between left/occipital-right/central regions and left/central-right/temporal regions. CONCLUSIONS: EEG beta coherence and especially the degree of generalized synchronization between a few inter-hemispheric regions during resting state with eyes closed allow a high accuracy classification rate of ADHD subjects. SIGNIFICANCE: Changes in inter-hemispheric EEG functional brain connectivity at rest are useful for ADHD diagnosis in a broad age range.

Assessment of electroencephalographic functional connectivity in term and preterm neonates.

OBJECTIVE: To study how functional connectivity of neonate EEG during sleep is assessed by different interdependence indices and to analyze its dependence on conceptional (CA), gestational (GA) and/or chronological age (CRA). METHODS: EEG data from eight cortical regions were recorded during active (AS) and quiet sleep (QS) in three groups of seven neonates each: preterm (PT; GA: 33-34 weeks; CA: 39-40 weeks), junior-term (JT; GA: 38-39 weeks; CA: 39-40 weeks) and senior-term neonates (ST; GA: 38-39 weeks; CA: 44-45 weeks). EEG functional connectivity was assessed by means of the coherence function (its magnitude (MSC) and its imaginary part (IMC)) and a measure of phase synchronization called phase lag index (PLI). RESULTS: Inter-hemispheric connectivity: (a) during AS in the beta band, the MSC of the ST group was greater than that of the PT group for the temporal region; (b) during QS in the delta band, both PLI and IMC of the ST group were different to those of the PT group for the frontopolar and central regions, whereas ST-JT differences were only found for PLI. Intra-hemispheric connectivity: (a) during AS in the beta band the MSC of the ST group was greater than that of the PT group for the left frontopolar-centrotemporal and right occipital-centrotemporal regions; (b) during QS in the beta band, both IMC and PLI were different for the JT group than for the PT and the ST groups for the right and left occipital-centrotemporal regions. CONCLUSIONS: EEG inter- and intra-hemispheric functional connectivity in neonates during sleep changes with the CA and CRA in delta and beta bands. SIGNIFICANCE: The neonate's brain development during the first weeks of life can be traced from changes in the characteristics of EEG functional connectivity.