Machine Learning-Based Electroencephalographic Phenotypes of Schizophrenia and Major Depressive Disorder

Background: Psychiatric diagnosis is formulated by symptomatic classification; disease-specific neurophysiological phenotyping could help with its fundamental treatment. Here, we investigated brain phenotyping in patients with schizophrenia (SZ) and major depressive disorder (MDD) by using electroencephalography (EEG) and conducted machine-learning-based classification of the two diseases by using EEG components.Materials and Methods: We enrolled healthy controls (HCs) (n = 30) and patients with SZ (n = 34) and MDD (n = 33). An auditory P300 (AP300) task was performed, and the N1 and P3 components were extracted. Two-group classification was conducted using linear discriminant analysis (LDA) and support vector machine (SVM) classifiers. Positive and negative symptoms and depression and/or anxiety symptoms were evaluated.Results: Considering both the results of statistical comparisons and machine learning-based classifications, patients and HCs showed significant differences in AP300, with SZ and MDD showing lower N1 and P3 than HCs. In the sum of amplitudes and cortical sources, the findings for LDA with classification accuracy (SZ vs. HCs: 71.31%, MDD vs. HCs: 74.55%), sensitivity (SZ vs. HCs: 77.67%, MDD vs. HCs: 79.00%), and specificity (SZ vs. HCs: 64.00%, MDD vs. HCs: 69.67%) supported these results. The SVM classifier showed reasonable scores between SZ and HCs and/or MDD and HCs. The comparison between SZ and MDD showed low classification accuracy (59.71%), sensitivity (65.08%), and specificity (54.83%).Conclusions: Patients with SZ and MDD showed deficiencies in N1 and P3 components in the sum of amplitudes and cortical sources, indicating attentional dysfunction in both early and late sensory/cognitive gating input. The LDA and SVM classifiers in the AP300 are useful to distinguish patients with SZ and HCs and/or MDD and HCs.

Deconstructing the Cortical Sources of Frequency Following Responses to Speech: A Cross-species Approach

Time-varying pitch is a vital cue in the processing of speech signals. Neural processing of time-varying pitch cues in speech has been extensively assayed using scalp-recorded frequency-following responses (FFRs), which are thought to reflect integrated phase-locked activity from neural ensembles exclusively along the subcortical auditory pathway. Emerging evidence however suggests that the auditory cortex contributes to the FFRs as well. However, the response properties and the relative cortical contribution to the scalp-recorded FFR are only beginning to be explored. Here we used direct intracortical recordings from human subjects and animal models (macaque, guinea pig) to deconstruct the cortical sources of FFRs and leveraged representational similarity analysis as a translational bridge to characterize similarities between the human and animal models. We found robust FFRs in the auditory cortex that emerged from the thalamorecepient layers of the auditory cortex and contributed to the scalp-recorded FFRs via volume conduction.

Tensor Decomposition of Human Oscillatory EEG Activity in Frequency, Space and Time

Changing physiological conditions in the central nervous system are associated with excitation and inhibition of cortical neuronal sources, many of which are reflected in modulation of narrowband scalp‐ recorded EEG oscillations (NSEOs). NSEOs exhibit specific electric field patterns on the scalp, which are largely determined by the geometry of the underlying cortical sources. Isolating NSEOs using spectral and spatial filters has led to many useful applications, from understanding mechanisms of drug action, to deeper understanding of sensory, perceptual, and cognitive functions. However, the scalp-recorded EEG combines signals of multiple NSEOs and massively distributed broadband cortical sources, which in turn greatly limits the practical utility of NSEOs.Over the past 10 years we have been developing methods to improve the measurement of NSEOs using tensor decompositions such as parallel factor analysis (PARAFAC). We and others have shown that PARAFAC can accurately model NSEO activity as a tensor product of dimensions of frequency, space and time. We introduced frequency and spatial constraints, which have improved the physiological plausibility of the NSEO models. In this paper we demonstrate the principle of the tensor approach using simulated scalp EEG data obtained by forward modeling. This allows us to carefully manipulate the spectral, spatial and temporal attributes of NSEOs and validate the obtained solutions. We observe superior performance of the tensor approach when compared with spatio-spectral decomposition, a broadly used technique for measuring oscillatory activity. This is achieved without a priori narrowband filtering, which is inappropriate when isolating and measuring NSEOs with unknown spectral properties.