Computational protocol: Assessing the feasibility of online SSVEP decoding in human walking using a consumer EEG headset

Similar protocols

Protocol publication

[…] This study employed a 14-channel Emotiv EEG headset that sampled EEG signals at 128 Hz and band-pass filtered them between 0.2 and 45 Hz. The electrodes were positioned at AF3, F7, F3, FC5, T7, P7, O1, O2, P8, T8, FC6, F4, F8, and AF4 in accordance with the modified international 10–20 system. In addition, the angular velocities were concurrently recorded by the built-in two-channel gyroscope (Gyro-X: horizontal movement; Gyro-Y: vertical movement) in order to characterize the severity of the head movements during walking. The Emotiv Control Panel software provides visual monitoring of the electrode impedance. Once all electrodes had good contacts with the scalp, a laptop (Thinkpad X230, Lenovo Inc.) started to initiate the OpenViBE software [] for connecting to the Emotiv headset, and then activate the BCILAB toolbox [] to stream EEG signals for real-time frequency detection of the SSVEP signals. [...] The offline analyses aimed to explore the dynamics of the background EEG activity and the SSVEP amplitude acquired by the consumer headset under different walking speeds. The background EEG spectrum in the fixation-cross condition, which can be regarded as baseline power for the SSVEP conditions under each walking speed, was necessary to reveal the underlying brain activity associated with the walking locomotion. In addition, since this study adopted treadmill walking to systematically test the effects of different degrees of head movements on the SSVEPs, it is important to report the intensity and frequency of the head movements. To this end, we applied the spectrum analysis to the 2-channel gyroscope signals (horizontal and vertical axes) along different walking speeds. The detailed steps are described as follows. This study applied a 1 Hz high-pass finite impulse response (FIR) filter to the EEG signals to remove low-frequency drifts, resulting in a signal bandwidth of 1 to 45 Hz for the spectral analysis. The 128-point short-time Fourier transform (STFT) with a Hamming window of length 128 samples and 25% overlap was then applied to estimate the EEG spectrogram at a 1 Hz frequency resolution. The grand average power spectral density (PSD) of each channel can be derived for each condition. Note that, to dissociate the reactive SSVEP from the background EEG activities and walking related noises, this study estimated the relative power of SSVEPs by subtracting the power spectrum in the fixation-cross condition from those of the SSVEP conditions at each of the walking speeds. In addition, this study calculated the scalp distributions of the spectral power to form topographic maps using the EEGLAB toolbox []. The spectrogram of the recorded 2-channel Gyro signals was estimated using the same procedure as for EEG data (but without filtering). Lastly, this study employed a paired t-test to access the differences of SSVEP amplitudes and online performance between two walking conditions (standing, 0.45 m/s, 0.89 m/s, and 1.34 m/s). […]

Pipeline specifications

Software tools BCILAB, EEGLAB
Application Clinical electrophysiology
Organisms Homo sapiens