Brain functional connectivity associates with fatigue in SSVEP-BCI applications

Abstract

Introduction: Fatigue is considered as a major challenge of practical use of steady state visual evoked potential (SSVEP) based brain-computer interfaces (BCIs) as it brings hazard to health and degradation of system performance [1]. In this work, we performed EEG based brain network analysis to study the fatigue development in SSVEP-BCIs applications. An experiment has been designed to induce fatigue to 2 subjects during an SSVEP-BCI task. The connectivity in frequency bands of delta, theta, alpha and beta based on magnitude square coherence of EEG signals between electrodes has been calculated to build the functional network. The network parameters have been investigated according to fatigue references based on subjective self-assessment and system performance. Material and Methods: The experiment protocol is concisely presented in the sub-figure a: The subjective fatigue evaluation is conducted with Stanford Sleepiness Scale. It is a well-accepted and very concise fatigue scale, so that the recurrent fatigue assessment would not be too disturbing to the subjects. The threshold for connectivity was chosen individually to deal with the individual difference [2-4]. Results: Between pre- and post- baseline measurement, delta and alpha bands connectivity show most active changes. The connectivity in delta band decreased in right frontal and increased in left central and occipital area while alpha band connectivity increased in right frontal area in eye-closed baseline measurement. The connectivity between pre- and post open-eyes baseline did not show consistent results across subjects. In SSVEP task sessions, alpha and beta bands connectivity show most active changes. Active changes happen in frontal, central and occipital areas. As subjects' fatigue level increase, the alpha and beta bands connectivity within these areas decrease; when subjects refreshed during the rest intervals, the connectivity increased. Some results of changes from both subjects are shown in the sub-figure b. Discussion: The changes in frontal area in right frontal agreed with studies [2], which reflected the changes in attentions between alert and fatigue states in baseline measurement. The changes in central and occipital areas during the SSVEP tasks are highly possibly due to the increasing workload in visual cortices and the drowsiness of subjects accumulated during the experiment. Significance: Investigated the changes of brain dynamic state in the fatigue development during SSVEP-BCI tasks. More information about the mechanism of fatigue in SSVEP-BCIs is explored. Acknowledgement: Supported in part by Macau Science and Technology Development Fund (036/2009/A, 142/2014/SB and 055/2015/A2) and Univ. of Macau Research Committee (MYRG: 139-FST11-WF, 079-FST12-VMI, 069- FST13-WF, 2014-00174-FST, 2016-00240-FST and 2017-00207-FST).