High and low gamma EEG oscillations in central sensorimotor areas are conversely modulated during the human gait cycle
Graphical abstract
Introduction
The ability to walk safely and independently is important for humans. Cortical injuries (e.g., stroke) can cause motor impairment and lead to limitations in the execution of daily life activities. Thus, great effort is put into restoring walking in people with motor impairments. To get a deeper understanding of cortical involvement during walking it is necessary to develop models, which are capable of describing cortical activities in relation to human walking patterns. Such models could facilitate the development of novel rehabilitation strategies in the future.
Neuroimaging studies using functional magnet resonance imaging (fMRI) restrict subjects to a lying position with fixated heads. Therefore, such setups are not well-suited for studying human brain function during walking. To overcome these methodical limitations electroencephalographic (EEG) source imaging (Baillet et al., 2001, Michel et al., 2004) can be used. Despite its low spatial resolution (centimeters), sophisticated analysis of the EEG offers several advantages. First, the temporal resolution of EEG signals in milliseconds allows analyzing cortical processes in relation to walking patterns. Second, analysis in the frequency domain opens possibilities to investigate different elements of cortical activity (Buzsáki and Draguhn, 2004, Siegel et al., 2012). Third and most important for investigating cortical involvement during walking, EEG source imaging can be done in ambulatory conditions (i.e., mobile brain imaging).
In recent years, several studies have investigated brain activity during walking (Gwin et al., 2011, Gramann et al., 2010, Presacco et al., 2011, Severens et al., 2012, Petersen et al., 2012, Wagner et al., 2012, Wagner et al., 2014, De Sanctis et al., 2014, Ehinger et al., 2014, Lau et al., 2014, Seeber et al., 2014). In agreement with earlier studies of isolated foot movement (Pfurtscheller et al., 1997, Crone et al., 1998, Miller et al., 2007, Müller-Putz et al., 2007), β oscillations in central sensorimotor areas were found to be suppressed (event-related desynchronization, ERD) during walking relative to a non-movement reference (Wagner et al., 2012, Severens et al., 2012, Seeber et al., 2014). Additionally, low γ (25–40 Hz) amplitudes were found to be modulated locked to the gait cycle (Wagner et al., 2012, Wagner et al., 2014, Seeber et al., 2014). The same frequency range was reported by Petersen et al. (2012) for significant coherence between EEG recordings over leg motor areas and the anterior tibialis muscle.
Sustained β suppression and low γ modulation were found to be simultaneously present and superimposed in the frequency domain and in spatial location during walking. Nevertheless, the different spectral peaks of β suppression and low γ modulation suggest that these phenomena are different elements of EEG activity during walking. We proposed that altered levels of β suppression during walking signify enhanced cortical excitability in central sensorimotor areas. Furthermore, gait cycle related modulation of low γ amplitudes may reflect sensorimotor processing linked to the motion sequences (Seeber et al., 2014).
In this work, we further develop the electrophysiological model of walking, including data from higher frequency oscillations (> 50 Hz). Previous studies showed high γ oscillations (60–90 Hz) to play an important role in motor execution (Crone et al., 1998, Pfurtscheller et al., 2003, Miller et al., 2007, Cheyne et al., 2008, Ball et al., 2008, Donner et al., 2009, Muthukumaraswamy, 2010, Darvas et al., 2010, Joundi et al., 2012). High γ power increase in electrocorticographic (ECoG) recordings correspond spatially well to fMRI activity (Hermes et al., 2012a) and its superior focal distribution enables the decoding of single finger movement (Kubánek et al., 2009, Miller et al., 2009, Scherer et al., 2009, Hermes et al., 2012b). The feasibility of detecting high γ activity in the motor system from non-invasive recordings was reported for magnetoencephalography (MEG) (Cheyne et al., 2008, Dalal et al., 2008, Donner et al., 2009, Muthukumaraswamy, 2010) and EEG (Ball et al., 2008, Darvas et al., 2010) during isolated limb movements. However, due to muscular [electromyographic (EMG)] and movement artifacts, it is very challenging to detect high γ activity from EEG recordings during walking. EMG artifacts during body movements affect EEG recordings in a wide range of frequencies (~ 20–300 Hz) (Muthukumaraswamy, 2013, Castermans et al., 2014).
Extending the previous findings of our group (Wagner et al., 2012, Seeber et al., 2014) we focus on two different aspects of the walking experiment: the fact that a person walks and the rhythmicity of walking movements. Therefore, we first investigate differences of the amplitude spectra between conditions walking and standing. In these analyses we introduce a novel artifact correction method based on spectral decomposition to minimize the impact of muscular influence on EEG source images. This correction method enables us to analyze high γ activity during walking. Second, we examine amplitude modulations in relation to gait phases reflecting the rhythmicity of walking movements.
Section snippets
Experiment and recordings
Data were taken from a previous study of our group (Wagner et al., 2012). Ten healthy volunteers (5 female, 5 male, 25.6 ± 3.5 years) completed four runs (6 min each) of active walking and three runs of upright standing (3 min each) in a robotic gait orthosis (Lokomat, Hocoma, Switzerland). Walking speed was constant and adjusted for each participant individually ranging from 1.8 to 2.2 km per hour. The Lokomat was operated with 100% guidance force and body weight support was less than 30% in every
Muscular artifact correction
The spatial map of the first PSC (with the largest eigenvalue) showed activity located in lateral and dorsal regions close to the location of head and neck muscles (Figs. 1a, 2c). The eigenvalue of the first PSC was 5–10 times bigger than the eigenvalue of the 2nd one in every subject (Fig. 2b). Moreover, the spectral profile of this component increases from 2–20 Hz and remains at a certain level for higher frequencies (Figs. 1b/c, 2a/c, S1). The spatial location and spectral profile of the
Muscular artifact correction
The spatial location and the frequency spectrum of the first principal spectral component (Figs. 1a–c) suggest this component to represent muscular artifacts. The first PSC can be robustly identified by its eigenvalue, because its magnitude is 5 to 10 times larger than the eigenvalue of the 2nd PSC in every subject (Fig. 2b). Spatially widespread activity and such with high amplitudes lead to large eigenvalues in our approach. Both criteria are given for muscular activities and therefore
Acknowledgments
This work was supported by the European Union research project BETTER (ICT-2009.7.2-247935) (www.car.upm-csic.es/bioingenieria/better/), BioTechMed Graz and the Land Steiermark projects BCI4REHAB (bci.tugraz.at/bci4rehab) and rE(EG)map! (bci.tugraz.at/reegmap). This is the sole opinion of the authors and funding agencies are not liable for any use that may be made of the information contained herein.
References (63)
- et al.
Movement related activity in the high gamma range of the human EEG
Neuroimage
(2008) - et al.
The functional role of cross-frequency coupling
Trends Cogn Sci.
(2010) - et al.
About the cortical origin of the low-delta and high-gamma rhythms observed in EEG signals during treadmill walking
Neurosci. Lett.
(2014) - et al.
Self-paced movements induce high-frequency gamma oscillations in primary motor cortex
Neuroimage
(2008) - et al.
Five-dimensional neuroimaging: localization of the time–frequency dynamics of cortical activity
(2008) - et al.
Cortical surface-based analysis. I. Segmentation and surface reconstruction
Neuroimage
(1999) - et al.
High gamma mapping using EEG
Neuroimage
(2010) - et al.
Recalibration of inhibitory control systems during walking-related dual-task interference: a Mobile Brain-Body Imaging (MOBI) study
Neuroimage
(2014) - et al.
A framework for local cortical oscillation patterns
Trends Cogn. Sci.
(2011) - et al.
Buildup of choice-predictive activity in human motor cortex during perceptual decision making
Curr. Biol.
(2009)