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

Abstract

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.

Functional Connectivity Analysis in Motor-Imagery Brain Computer Interfaces

“…Given the hypothesis-based characteristics of this study, we opted for phase synchronization analysis using PLV, which does not involve the magnitude of the signals but rather the phase of the signals (Sakkalis, 2011). Additionally, in order to reduce the susceptibility of our analysis to volume conduction artifacts, we applied a Laplacian spatial filter to the EEG recordings (Cohen, 2014;Kayser and Tenke, 2015;Caicedo-Acosta et al, 2021). The problem of volume conduction, however, is still prevalent in PLV analyses and further refinement of phase synchrony estimation algorithms that are cost-effective and robust to volume conduction remains necessary (Bruña et al, 2018).…”

“…In the former studies, both effective connectivity (e.g., Lee et al, 2020) and functional connectivity (e.g., Vidaurre et al, 2020) measures have been investigated, however, these studies employed various metrics of connectivity including coherence, phase synchronization, phase-slope index, etc., which employ different algorithms and hence vary in their interpretation (Bastos and Schoffelen, 2016). However, to fully tackle the disadvantages of EEG, such as artifacts and inter-trial/inter-subject amplitude variability, phase-based relationships (e.g., phase synchronization) might provide the best functional connectivity measure of spatially distributed regions that are active during mental task execution (Caicedo-Acosta et al, 2021). Functional connectivity features measured by the phase lag index (PLI) and phase-locking value (PLV) can discriminate between different MI tasks (Stefano Filho et al, 2018;Caicedo-Acosta et al, 2021), and therefore are a promising tool to identify potential non-learners (Caicedo-Acosta et al, 2021).…”

“…However, to fully tackle the disadvantages of EEG, such as artifacts and inter-trial/inter-subject amplitude variability, phase-based relationships (e.g., phase synchronization) might provide the best functional connectivity measure of spatially distributed regions that are active during mental task execution (Caicedo-Acosta et al, 2021). Functional connectivity features measured by the phase lag index (PLI) and phase-locking value (PLV) can discriminate between different MI tasks (Stefano Filho et al, 2018;Caicedo-Acosta et al, 2021), and therefore are a promising tool to identify potential non-learners (Caicedo-Acosta et al, 2021).…”

“…By comparing two groups of High and Low BCI performers in a large dataset (N = 54), this study intends to investigate if successful MI task execution (i.e., better BCI performance) is associated with establishing a different (perhaps stronger) connectivity pattern between brain areas. As phase synchronization is a promising measure in predicting MI-BCI performance, which is also robust to artifact and inter-subject amplitude variability (Caicedo-Acosta et al, 2021), PLV is employed as a measure of functional connectivity in this study.…”

Makine Öğrenmesi Kullanarak Krom Kaplama Örtme Gücünün Tahmin Edilmesi

“…Makine öğrenmesi metotları son yıllarda mühendislik alanlarında yaygın olarak kullanılmaktadır [25][26][27]. Hem regresyon hem de sınıflandırma verilerinde tercih edilmektedir.…”