Research On Recognizing The Intention Of Walking And Gait Regulation Based On Cerebral Hemoglobin Information

Faced with increasingly aging,stroke and other problems,more people in contemporary society have suffered from lower limb motor dysfunction.It is of great social significance to assist patients with lower limb motor dysfunction to walk independently,which can reduce the burden on caregivers.Recently,walking assisted devices based on sound,button and other traditional control methods are not suitable for the elderly or stroke patients with speech barrier,weak limb operation ability and poor cognitive ability.Therefore,the brain-computer interface(BCI)technology is used to identify the spontaneous motion intention of the human body to control the walking assisted devices to solve this problem.However,current BCI researches on the recognition of human motion intention is still limited to the recognition of discrete motion state,which deviates from the continuous motion in daily life.And these studies only focus on off-line recognition of motion state,lack of consideration of real-time performance and online verification,which lead to the limitation of brain-computer interface technology in practical application.Therefore,in this paper,the near-infrared spectroscopy brain-computer interface technology(NIRS-BCI)is used to identify the walking intention in real time in order to control the walking assisted devices.Meanwhile,in order to improve the adaptability of patient control equipment and the practical application value of recognition algorithm,the continuous state discrimination method of gait-regulated intention is also studied.The main research contents and methods are as follows:(1)Four kinds of gait-regulated walking experiments were designed,and 30 volunteers were recruited to participate in the experiment.The brain hemoglobin information was collected by near-infrared spectroscopy when the volunteers carried out the walking experiments.(2)The recognition method of walking intention was studied.In order to meet the real-time requirements,a fast feature extraction method based on Teager-Kaiser energy operator was proposed.In the face of large amount of data,unbalanced modeling task,the decision model based on GBDT(Gradient Boosting Decision Tree)algorithm was adopted.(3)The recognition method of gait-regulated intention was studied.In order to filter environmental and physiological interference noise,a combined filtering method based on morphological filtering was proposed.According to the difference of skull size,a method of sub-brain division and blood oxygen concentration calculation based on entropy method was proposed.In order to meet the real-time requirements,the sliding window method was used to extract the features of time domain and correlation features.In the face of the multi-state recognition task of gait adjustment,a decision model based on Stacking algorithm integrating GA-SVM(Genetic Algorithm-Support Vector Machine)was proposed.(4)The offline recognition model of walking intention and gait-regulated intention were integrated.Then,the walking and gait-regulated intentions of 10 subjects was tested online and controlled external robots in real time by building an online verification platform of brain-computer interface.The online performance of the walking and gait-regulated intention are ACC(Accuracy)=100%and 70%,FR(False Rate)=2.91%and 0.70%and LAY(Detection Latency)=0.39±1.06 seconds and-0.71±2.01 seconds respectively.The execution time of the online recognition program is 36.99±4.20 milliseconds.This result demonstrates the feasibility of identifying walking and gait-regulated intentions based on cerebral hemoglobin information and controlling walking assisted devices.This study explored a new method to identify the dynamic regulated-intention based on BCI and realized the real-time discrimination of walking and four gait regulation intentions with higher discriminative accuracy.This research enhanced the practical application value of the recognition algorithm of motion intention based on brain information,improved the intelligence and adaptability of the walking assisted devices based on BCI,and made a preliminary attempt to apply brain-computer interface technology into the field of rehabilitation.