To Identify Motion Pattern of Lower Limbs by Using Cerebral Hemoglobin Information during Motor Imagery

To increase intelligence of walking-assistive devices, it is important to identify motion intention of the subject who wears a walking-assistive device. In this paper, cerebral hemoglobin information during motor imagery was used for recognizing frequently performed motion patterns of walking, squatting and standing up. 14 subjects performed spontaneous movements: the start and end of each task, and the time duration of every task were controlled by subjects themselves. Before an actual movement in each pattern, a corresponding motor imagery task was carried out. During the experiment, cerebral hemoglobin information was recorded by using near-infrared spectroscopic technology. Wavelet packets decomposition was performed on oxygenated hemoglobin (oxyHb) with a minimum resolution of 0.03Hz. As for the frequency bands of 0-0.03Hz, 0.03-0.06Hz, 0.06-0.09Hz, 0.09-0.12Hz, and 0.12-0.15Hz, the means and change rates of regional oxyHb during the motor imagery periods before actual movements were selected as feature vectors. Pattern identification was performed by using SVM method, and a recognition rate of 82.3% was achieved. Since the data before actual movements was used for recognizing motion patterns, the results will be preferable to send a control command timely for a walking-assistive device.