Back to results list
Show full item record
Please use this identifier to cite or link to this item:
|Title:||Error monitoring during gait training modulates theta band oscillation : an EEG study|
|Source:||The 22nd Annual Meeting of the Organization for Human Brain Mapping (OHBM), Geneva, Switzerland, 26-30 June 2016, poster number 1760 (Poster) How to cite?|
|Abstract:||Introduction: People acquire motor skills mostly from repeated error-based trainings. Previous studies discussed the neural mechanism of error monitoring in cognitive tasks , while how error message modulates neural process during lower-limb motor learning remains unknown. This study investigated how brain discriminates positive (PF) and negative (NF) visual feedbacks to modify gait mechanics. Peak positive acceleration (PPA) during walking, representing the landing impact , was the target parameter to be reduced in our training. Electroencephalography (EEG) was recorded to capture cortical dynamics.|
Methods: Methods 26 healthy subjects were recruited to complete two randomized walking sessions (1.25 m/s, 3 minutes). One minute of walking PPA was collected after 3-minute treadmill adaptation, and the corresponding mean value was regarded as the baseline of each subject's landing impact. In no-feedback sessions, subjects were asked to walk naturally on a treadmill. In visual feedback sessions, a visual signal time-locked to each right heel-on event was shown on a monitor to facilitate soft landing. A transient red signal (NF) was given if the corresponding PPA was over 80% of the baseline, indicating adjustments were needed, while a transient light green signal (PF) was given otherwise. A dark green signal was kept on the screen when no heel-on event was triggered (Fig 1). EEG channels were first screened based on their standard deviation (SD), kurtosis, and correlation with adjacent channels . Independent components (IC) and their equivalent current dipoles were then calculated in EEGLAB [4, 5, 6]. ICs representing artifact, or whose residual value exceeded 20% were excluded from further analysis . The remaining EEG data was segmented into epochs based on right heel-on events. The left and next right heel-on events corresponding to each epoch were time-warped to 50% and 100% of the gait cycle (GC) respectively. The onset of feedback was aligned to 10% of the GC. Event-related spectral perturbation (ERSP) was calculated for each epoch to show the intra-stride EEG oscillation in time-frequency domain. ICs were grouped into clusters using a k-means algorithm based on the feature vector consisting of their dipole location, power spectra and scalp map. To guarantee equal contribution of each subject to the cluster-wise analysis, only the IC that was closest to the cluster centroid was remained for each subject. A permutation resampling method was applied to determine significant gait ERSP for each cluster (p<0.05). For statistical analysis, theta-band (4–8 Hz) ERSP of each IC was averaged within each 5% GC, generating a 20-point ERSP curve . A paired t-test with false discovery rate correction was then conducted to compare the theta-band ERSP curve of each cluster between PF and NF conditions.
Results: Clusters with significant difference in theta-band ERSP between PF and NF conditions were observed in anterior cingulate cortex (ACC), extrastriate cortex (EC) and posterior parietal cortex (PPC) (Fig 2), while their EEG oscillation in other frequency bands were mostly independent of feedback type (Fig 3). In ACC, event-related synchronization (ERS) was observed at 65% - 80% of the GC (p<0.006) when NF was displayed. In EC and PPC, the NF-induced ERS appeared relatively earlier, at 30% - 45% (p<0.007) and 35% - 40% (p=0.002) of the GC respectively (Fig 4).
Conclusions: We observed active involvement of ACC, EC and PPC in discriminating PF and NF during gait training. EC plays an important role in acquiring visual signal , while PPC integrates multisensory information and directs it to the frontal region . ACC was shown to be intensively engaged in error correction and monitoring . The role of these brain regions, supplemented with the latencies of their theta-band ERS observed in this study, implies a possible neural network comprising EC, PPC and ACC, whose behavioral pattern may be modeled as reception-integration-correction.
|Appears in Collections:||Conference Paper|
Show full item record
Checked on Aug 13, 2017
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.