Acquisizione di segnali elettroencefalografici durante il cammino mediante una wireless body sensor network: uno studio di fattibilità = Detection of electroencephalographic signals during gait through a wireless body sensor network: a feasibility study

The interest in understanding the cortical involvement in the control of human gait has led to increasingly in-depth studies of sensorimotor integration during dynamic activities. However, the nature of corticospinal interactions underlying this mechanism remains still poorly understood due to technological limitations in acquiring electrophysiological (i.e. Electroencephalograms-EEG, Electromyograms-EMG) and biomechanical signals during dynamic tasks. Despite significant technological advancements over the past years, one of the main issues concerns the physical constraints associated with the wired technology of EEG devices, which makes the experimental setups bulky, prone to movement artifacts, thus unsuitable for dynamic tasks. In this context, this study tested an innovative wireless body sensor network to simultaneously collect EEG, sEMG, kinematic data and ground reaction forces. Twelve participants were recruited for the study to perform overground walking at two different speeds and jogging. Measurements were performed at the Faculty of Sport and Health Sciences of the University of Jyväskylä, Finland. EEG were recorded with a 32 electrodes cap, sEMG signals were detected from Tibialis anterior, soleus and gastrocnemius. Acceleration from the foot and ground reaction forces from both feet were also acquired during the tasks. Synchronous data acquisition was guaranteed by a wireless synchronization system (synch delay<0.5ms). Physiological and non-physiological artifacts corrupting EEG traces were distinguished from the continuous data. Non-physiological artifacts related to the gait movements were identified and classified as motion artifacts by analyzing the correlation between EEG and kinematic data (i.e. acceleration and ground reaction forces). Different approaches were applied to attenuate motion artifacts: Independent Component Analysis (ICA), Common Average Reference (CAR) and adaptive Common Average Reference (a-CAR). Three versions of the latter filter were implemented. Specifically, a-CAR was applied to: (i) continuous EEG signals, (ii) a 1-s sliding window of the signal, (iii) signal windows synchronous with each heel strike time instant. Finally, filtered EEG signals in the beta frequency band (12-35 Hz) were used to assess the corticomuscular coupling (EEG-EMG) during overground gait. Two main types of motion artifacts were identified and characterized: exploring and reference electrode related. The former artifacts as well as the physiological ones (e.g., eye blink artifacts) were removed through ICA. Regarding the artifacts arising from the reference electrode contact, although group analysis did not show significant differences between CAR and a-CAR, the latter seemed to have better performances in cases where artifact amplitude varied across EEG channels. Due to the high intra- and inter- subjects’ variability in artifact characteristics, it was not possible to find a common trend among the performances of the three a-CAR methods in terms of motion artifacts removal. Corticomuscular coupling analysis showed significant coherence between the medial gastrocnemius muscle and EEG of the electrodes located over the leg motor area. In conclusion, this study presents an overview of the current challenges in acquiring EEG during dynamic tasks and shows the feasibility of sensorimotor integration assessment during gait using an innovative wireless body sensors network for the recording of electrophysiological, kinematic, and biomechanical signals.