A Research On Classifying Two-dimensional Motion States Of Step Length And Walking Speed By Applying Cerebral Hemoglobin Information

For some patients who have severe motor dysfunction caused by the aging,stroke,spinal cord injury,and accidental injury,etc.Their patients' brain function may be degraded because of bedridden and no exercise for extended periods of time.Therefore,it's of great social significance to fully mobilize patients' consciousness of active rehabilitation training to carry out useful rehabilitation training,so as to help them recover the ability to walk independently.At present,the common method,controlling the external equipment to do rehabilitation training through the identified motion intention,is based on the patients' electromagnetic signals and biomechanical signals.However,it would make a great error,because of the atrophy and loss of patients' muscle.So the brain information can be used to judge the motion intention of patients.At present,the techniques used in the acquisition of brain information include functional magnetic resonance imaging,electroencephalograph,and functional near-infrared spectroscopy and so on.Moreover,the functional near-infrared spectroscopy compared with other technologies supports the continuous test of large amplitude motion in normal environment,and has a good time and spatial resolution.Therefore,this study proposes a method of controlling the external equipment through the identified motion intention by using functional near-infrared spectroscopy in real time.The main research methods in this paper are as follows:(1)Thirty-one healthy subjects were recruited to walk under six given sets of gait parameters,they are small-step with low/mid-speed,mid-step with low/mid/high-speed,and large-step with mid-speed,all the movements were controlled by themselves.(2)At the beginning of this experiment,all subjects were asked to take a 40 s rest,then,the maximum values of the standard deviation of different parameters in this period were calculated by the method of sliding window,respectively,which were defined as classifying two-dimensional motion states of step length and walking speed by applying cerebral hemoglobin information the criterions for judging the starting movement.And the next collected data was used to compare with its criterion,so as to confirm the starting position.Finally,the recognition accuracy of the starting position was 89.4%,and the rate of miscarriage in the rest time was only 1.23%.(3)The six gait parameters were classified under the different levels of step length and walking speed.The total hemoglobin and the difference between the oxygenated and deoxygenated hemoglobin signal of 1s before motions were used to classify different states.Wavelet packet decomposition was used to seek the key important channels and its corresponding important key regions of the brain.In order to eliminate the difference in the size of the head,typical feature vectors were represented by some flexible channels in predefined regions,and best combination of feature vectors was selected by applying a genetic algorithm and a library for support vector machine algorithm.For different step lengths,the walking speed recognition rate was found to be 71.21%,and for different walking speeds,the step length recognition rate was 71.21%.(4)Based on the above methods of judging the starting position and classifying different states,a real time verification platform for brain-computer interface was built to verify the feasibility of controlling the external robot through the identified brain signals.Another five subjects were recruited to carry out final testing,finally,the rate of miscarriage in the rest time was 5.7%,and for the condition of different step lengths level,the recognition rate was 66.7%,and for the condition of different walking speeds level,the recognition rate was 60%.Finally,it was proved that the method,controlling the external equipment through the identified motion intention by using functional near-infrared spectroscopy in real time this study proposed,was feasible according to the result of the experiment.This study realizes the idea of controlling the external robot in the real time by using the identified states of subjects' motion intention.It lays a foundation for controlling the external equipment to carry out rehabilitation training through patients' hemoglobin signals,so as to help patients recover their ability to walk independently and reduce social burden.