Research On Predicting Method Of Muscle Movement Parameters Based On Ultrasound

Skeletal muscle is an important tissue of the human body's composition.To understand the role of the skeletal muscle in body motions will be helpful for resolving some potential problems about musculoskeletal system in different human groups.At present,researchers have put forward many kinds of human-machine interface(HMI)to research musculoskeletal system.However,these HMIs have not been perfect.As a non-invasive,real-time and convenient imaging technology,ultrasonography has been increasingly used in the research on musculoskeletal system.This paper focuses on the application of ultrasound imaging in the prediction of musculoskeletal movement parameters.Firstly,we use the ultrasonic device to verify the ability of the ultrasonic HMI to identify the discrete gestures of the human body.In this paper,we constructed a database which includes 29 discrete actions of 15 subjects,and then extracted the ultrasound data feature and classified the gestures.Through different combination of the feature and the classifier,we verified the ultrasonic HMI prediction ability on the discrete motion.The results show that the classification accuracy of 29 discrete gestures can reach about 90%.Secondly,we use ultrasonic device to verify the prediction ability of ultrasonic HMI on the continuous movement of the human body.Most of the traditional methods are carried out by the neural network or the musculoskeletal model.The former method can not reveal the physiological process of motion,and the latter is complex.Based on this,this paper presents a physiologically significant model.Firstly,we extracted two anatomical features from the ultrasound video: muscle thickness and muscle contraction velocity,and then we established the equation between muscle thickness with joint continuous motion by combining with the muscle activation model and the muscle contraction model and other physiological models,and finally,in order to reduce the data noise,The muscle thickness and muscle contraction speed parameters were used to correct the equation as measurement value.In this way we got a closed-loop prediction mode,which can solve the limitations of the above two methods.On the basis of the proposed model,this paper has carried on the prediction experiment to the joint angle and the angular velocity to verify the prediction ability of the model.The final result proves that the model can predict the joint angle(correlation: 0.96,RMSE: 0.16)and joint angular velocity(correlation: 0.91,RMSE: 0.19)well.