Research And Design Of Virtual Reality Training System In Motor Function Rehabilitation For Stroke Patients

During stroke,the brain cannot supply blood normally,thus damaging some part of brain tissue and resulting in dyskinesia such as hemiplegia.Motor rehabilitation and regular assessment are needed in the early recovery stage of stroke.Compared with traditional motor rehabilitation,virtual reality motor rehabilitation system not only provides repeated and standard motor rehabilitation guidance for patients,but also records motor data and assess the recovery of patients’ motor ability.The main contents of this study are divided into two aspects: the design of virtual reality motor training system and the motion assessment.The hardware of the virtual reality motor training system is composed of 9-axis sensor MPU9250 and Wi-Fi transmission module.The 9-axis sensor,which consists of a 3-axis accelerometer,a 3-axis gyroscope and a 3-axis compass,transmits motion data to the virtual reality motor training system through Wi-Fi.The function of virtual reality motor training system includes software interface,wireless transmission and training mode.The sensor is worn at the target training joints of patients,which include shoulder joint,elbow joint and wrist joint.The training mode is divided into active mode and following mode.In the active mode,patients are asked to do rehabilitation actively,while the system feedback the movement based on collected data.In the following mode,the system offers a series of standardized Brunnstrom upper limb training instructions,while patients are asked to follow and do the same movement.During the rehabilitation,the sensors collect the current posture of patients in real time and get the attitude angle through data processing.The data is transmitted to user displayer through WiFi.In the user displayer,the posture angle is calculated and feedback to the virtual reality training system.The virtual reality scene displays the user’s joint angle,while the virtual character feedback the angle by doing the same movements.Patients can achieve immersive training through virtual reality rehabilitation.The motion assessment part selects 3 assessment items of Fugl-Meyer Assessment Upper End,which are Hand to Lumbar Spine,Shoulder Flexion 90 Degrees and Pronation-supination.Each item has 3 different rating levels with a total of 9 different action sets.The experiment collected 900 groups of motion data of 10 subjects for motion detection and motion recognition.The motion detection chooses short-term energy and zero-crossing rate method to analyze the starting and stopping points of the motion and then cut out a whole motion for motion recognition.The motion recognition uses 27 features from the 9-axis data for identifying 3 different motion from same item.The motion assessment part can effectively identify the patient’s movements.In the movement set of the experiment,the average accuracy rate of recognizing 3 different motion from same item is 98.68%.