Integrating mobile-EEG with a naturalistic extended reality task

Abstract

Naturalistic tasks can enhance the ecological validity of neuroscience research, particularly facilitate unravelling cognitive brain functions underlying developmental disorders where the symptoms are defined based on everyday life problems. Emergence of mobile electroencephalography (mEEG) allows more naturalistic tasks in EEG studies by enabling study paradigms that permit full body movement. However, movement causes notable artifacts into EEG data that require denoising. This thesis evaluated denoising methods and preprocessing pipelines for mEEG data in naturalistic extended-reality (XR) settings involving naturalistic whole body movement movement. The main objective of the present study was to develop a preprocessing pipeline to denoise mEEG data collected in a naturalistic paradigm. Twenty adults participated in the study, including 14 neurotypical (NT) and 6 diagnosed with attention-deficit/hyperactivity disorder (ADHD). Main methods for denoising the data included high-pass filtering at 2 Hz, inclusion of electromyography (EMG) sensors, and intercorrelating them with components derived by independent component analysis (ICA). Furthermore, other methods including filtering, EMG regression, Artifact Subspace Reconstruction (ASR) were compared. Results showed that high-pass filter at 2 Hz and inclusion of EMG-guided ICA improved denoising via making muscle-related ICs more distinguishable. Event-related potentials (ERPs) were calculated for distraction sounds and target-related responses for NT and ADHD adults to demonstrate the clinical potential of integrating mEEG with a naturalistic XR paradigm. Further studies with greater sample size should be conducted to confirm and expand the obtained findings.