Deep Neural Regression Prediction of Motor Imagery Skills Using EEG Functional Connectivity Indicators.

Motor imaging (MI) induces recovery and neuroplasticity in neurophysical regulation. However, a non-negligible portion of users presents insufficient coordination skills of sensorimotor cortex control. Assessments of the relationship between wakefulness and tasks states are conducted to foster neurophysiological and mechanistic interpretation in MI-related applications. Thus, to understand the organization of information processing, measures of functional connectivity are used. Also, models of neural network regression prediction are becoming popular, These intend to reduce the need for extracting features manually. However, predicting MI practicing's neurophysiological inefficiency raises several problems, like enhancing network regression performance because of the overfitting risk. Here, to increase the prediction performance, we develop a deep network regression model that includes three procedures: leave-one-out cross-validation combined with Monte Carlo dropout layers, subject clustering of MI inefficiency, and transfer learning between neighboring runs. Validation is performed using functional connectivity predictors extracted from two electroencephalographic databases acquired in conditions close to real MI applications (150 users), resulting in a high prediction of pretraining desynchronization and initial training synchronization with adequate physiological interpretability.

Non-stationary Group-Level Connectivity Analysis for Enhanced Interpretability of Oddball Tasks.

Neural responses of oddball tasks can be used as a physiological biomarker to evaluate the brain potential of information processing under the assumption that the differential contribution of deviant stimuli can be assessed accurately. Nevertheless, the non-stationarity of neural activity causes the brain networks to fluctuate hugely in time, deteriorating the estimation of pairwise synergies. To deal with the time variability of neural responses, we have developed a piecewise multi-subject analysis that is applied over a set of time intervals within the stationary assumption holds. To segment the whole stimulus-locked epoch into multiple temporal windows, we experimented with two approaches for piecewise segmentation of EEG recordings: a fixed time-window, at which the estimates of FC measures fulfill a given confidence level, and variable time-window, which is segmented at the change points of the time-varying classifier performance. Employing the weighted Phase Lock Index as a functional connectivity metric, we have presented the validation in a real-world EEG data, proving the effectiveness of variable time segmentation for connectivity extraction when combined with a supervised thresholding approach. Consequently, we performed a piecewise group-level analysis of electroencephalographic data that deals with non-stationary functional connectivity measures, evaluating more carefully the contribution of a link node-set in discriminating between the labeled oddball responses.