Fig. 5

EEG feature extraction maps. Top: Feature extraction maps obtained by a combination of CPCA and AIDA for classification of attempted walking versus idling. The spatial distribution of features is shown for the frequency bands centered at 15 Hz and 25 Hz, where the features with values close to ±1 are more important for classification. The maps were generated from data collected during the last visit. Bottom: Log power spectral density (PSD) during idling (blue) and walking (green) at electrodes CP3, CPz, and CP4, where shaded regions represent error bars. Underneath the PSD plots are the corresponding signal-to-noise ratio (SNR) plots with significant SNRs (p < 0.01) represented by red lines. Note the event-related synchronization (ERS) in the 13–16 Hz range (at CP3 and CP4) and event-related desynchronization (ERD) in the 23–28 Hz range (at CPz)