To monitor the limb movement two tri-axial accelerometers (ADXL353, Analog Devices, MA, United States) were used. One accelerometer was attached to the targeted limb contralateral to the stimulation site and the other was attached to another limb of interest. Comparing data from both accelerometers showed that we selectively stimulated the motor area controlling the targeted limb, while avoiding stimulation of the motor area controlling the other limb. The six axes (three from each accelerometer) were digitized (NI USB-6216, National Instruments) at 4 kHz sample rate, displayed and recorded for off-line analysis using custom written MATLAB software (MathWorks, MA, United States).
After the experiment, the raw acceleration data were band pass filtered between 3 and 500 Hz (second-order Butterworth) and integrated twice to give the limb displacement in arbitrary units. Principal component analysis was used to combine the three displacement axes and limb displacement was defined as the first principal component. The difference between the minimum and maximum limb displacement occurring after stimulation was then calculated to give limb displacement amplitude for each stimulus presentation.
After the experiment, the raw acceleration data were band pass filtered between 3 and 500 Hz (second-order Butterworth) and integrated twice to give the limb displacement in arbitrary units. Principal component analysis was used to combine the three displacement axes and limb displacement was defined as the first principal component. The difference between the minimum and maximum limb displacement occurring after stimulation was then calculated to give limb displacement amplitude for each stimulus presentation.