Research On Virtual Reality Rehabilitation For Forearm Amputees

There are numerous patients with forearm amputation in China.Wearing surface electromyography(sEMG)to control forearm prosthesis can improve the patient’s quality of life and catch up with the patient’s psychological trauma.Before the amputee installs the prosthesis,rehabilitation training of the remaining limbs can help promote blood circulation in the affected area,maintain the vitality of the nerves and muscles,help the patient adapt to the pyroelectric prosthesis control and improve the prosthesis control effect.At present,the physical rehabilitation training equipment is expensive.The training process is boring,and lacks visual and effective rehabilitation evaluation method.Gradually with the progress of science and technology,virtual reality technology has been applied in the field of rehabilitation medicine,so this topic is based on image recognition and analysis methods of electricity technology,puts forward a kind of virtual reality rehabilitation training system in forearm amputation patients,and then designs the effective rehabilitation evaluation methods,to enhance the fun and interactivity in the process of rehabilitation training.This project uses a binocular structured light camera(Realsense435i)to obtain a binary image of the upper limb residual limb contour by skin color segmentation method,and performs morphological operations such as expansion and corrosion to remove noise interference,uses the contour tracking method to obtain the main arm Contour,and then use Zhang’s fast parallel thinning algorithm to obtain the end point and inflection point of the subject’s forearm image.These points are clustered by K-means to obtain the endpoint coordinates of the two ends of the forearm.The depth information of the two ends is achieved by combining the depth images.Finally,the posture of the amputee’s forearm residual limb is successfully identified to control the spatial posture of the virtual hand in the virtual reality environment.At the same time,through the sEMG acquisition module(Myo armband),the sEMG signal(8 channels)of the amputee’s residual limb is obtained,and four timedomain features of average absolute value,zero-crossing rate,symbol slope change and waveform length are extracted.Support vector machine(SVM)classifier decodes sEMG information,recognizes the patient’s movement intention,and controls the virtual hand to achieve complex hand movements.Then,this project set up a simulated training scenario on the computer side,and designed a rehabilitation training program for the rehabilitation requirements of forearm amputation patients.Let the patient use the sEMG control virtual hand to place the virtual balls randomly scattered around the screen into the ball frame.During this process,the patient exercises the residual limb muscles and maintains muscle vitality through imaginary wrist extension and wrist flexion movements.At the same time,the shoulders and elbows are also stretched to increase joint mobility.Finally,three experiments are designed for this subject,namely,the experiment of collecting 2 movements,the experiment of collecting 5 different movements and the experiment of collecting 2 movements in 3 different positions.Five healthy subjects were recruited to wear gloves to simulate forearm amputees,and then the EMG signals of different hand movements and arm positions were collected.Finally,they participated in a virtual reality training task designed.The subjects faced the computer screen,operated the identified virtual hand of the stump,picked up the randomly appeared virtual balls in the screen,and put them into the fixed position of the ball frame.The offline classification accuracy of the EMG signals and the average total time for the subjects to complete the training task in real time were calculated by the virtual rehabilitation system.In the two-action experiment,five subjects spent an average of 57.72 seconds to complete the task,with an average offline accuracy of 89.03%.In the experiment,the two movements took an average of 54.98 seconds to complete the task in three different positions,with an average offline accuracy of 91.73%.In 5 experiments of different movements,the average recognition accuracy of the training system for a variety of hand movements reached 90%,which can be applied to the rehabilitation training of forearm amputees.In summary,the project is intended for the rehabilitation training of residual limbs of patients with forearm amputation.Based on image recognition and sEMG analysis technology,it provides a novel and effective virtual reality rehabilitation training method,which enhance the interactivity and rehabilitation of rehabilitation training.The effectiveness of the effect evaluation can help patients effectively train forearm residual limb muscle groups,and lay the groundwork for improving the control effect of prosthetic limbs.