基于体感交互技术的上肢主动康复系统的设计及信效度分析

Design of Upper Limb Active Rehabilitation System Based on Somatosensory Interaction Technology and Reliability and Validation Analysis

目的:将Kinect体感交互技术用于脑卒中导致的偏瘫患者上肢运动康复,结合脑卒中康复的基本理论及各康复阶段的特点,设计出针对上肢运动功能恢复的主动康复系统,以期提高康复效果。方法:通过3D建模技术创造虚拟训练场景,结合Kinect体感交互技术对人体姿势及动作进行识别,基于中枢神经系统的可塑性、Brunnstrom理论、运动再学习等康复理论,设计出符合脑卒中上肢运动功能障碍特点的虚拟游戏,同时实现患者即时训练信息的实时反馈,建立上肢运动康复训练系统,并验证该康复系统对上肢关节活动度测量的效度与信度。结果:应用体感交互技术设计出上肢主动康复系统,包括切食物、射击等多个虚拟训练游戏,通过运动检测及判别算法,实现患者与虚拟环境的交互,同时实现运动功能评估及关节活动信息的实时反馈。系统对肩关节外展、屈曲、后伸与肘关节屈曲等关节活动度测量与设定的目标值有较高的一致性,采用Bland-Altman进行比较,系统测量肩关节外展活动度的95%一致性界限为(-1.29,1.57),肩关节屈曲活动度的95%一致性界限为(-1.46,1.60),肩关节后伸活动度的95%一致性界限为(-1.53,1.67),肘关节屈曲关节活动度的95%一致性界限为(-1.42,1.56),系统效度良好。采用组内相关系数(ICC)评价系统的信度,其中在复测信度方面,肩关节外展、屈曲、后伸与肘关节屈曲等4个关节活动度的ICC均>0.9;研究者间信度方面,4个关节活动度的ICC均>0.9。结论:本研究设计的基于体感交互技术的上肢主动康复系统具有良好的效度与信度,可行性强,能满足临床上肢功能康复的需求,该系统为脑卒中后上肢功能康复提供新的方法,相较于传统的上肢训练具有针对性强、趣味性高等优势,有望提高康复治疗的效率。

Objective:To develop an upper limb active rehabilitation system with Kinect somatosensory interaction technology based on the basic rehabilitation theory and different stages of rehabilitation treatment for the post stroke hemiplegic patients to improve the effect of rehabilitation treatment.Methods:Virtual training scenarios were created with 3D modeling technology,and the human body posture and movements were detected with the Kinect somatosensory interaction technology.Based on the plasticity of the central nervous system,Brunnstrom theory,motor re-learning,and other rehabilitation theories,this study designed the virtual games in line with the characteristics of upper limb motor dysfunction after stroke.Meanwhile,the real-time feedback was given to the training of patients.Finally the upper limb exercise rehabilitation training system was established,and the validity and reliability of the rehabilitation system for the measurement of upper limb joint range of motion were verified.An upper limb active rehabilitation system was designed using the Kinect somatosensory interaction technology.Results:The system included several virtual trai-ning games such as cutting food and shooting.Through motion detection and discrimination algorithm,the system realized the interaction between patients and virtual environment,as well as the evaluation of motor function and real-time feedback of joint range of motion information.The measurement of shoulder abduction,flexion,extension,elbow flexion and other joint range of motion in the system had a high consistency with the set target values.The Bland-Altman was used for comparison.The 95%consistency limits of shoulder abduction range of motion in the system was(-1.29,1.57).The 95%consistency limits for shoulder flexion range of motion was(-1.46,1.60),the 95%consistency limits for shoulder extension range of motion was(-1.53,1.67),and the 95%consistency limits for elbow flexion range of motion was(-1.42,1.56),indicating good system validity.Intra-group correlation coefficient(ICC)was used to evaluate the reliability of the system.In test-retest reliability,the ICC of shoulder abduction,flexion,extension and elbow flexion were all greater than 0.9.In terms of interobserver reliability,the ICC of all four joint range of motion were greater than 0.9.Conclusion:This upper limb active rehabilitation system designed based on somatosensory interaction technology has good validity and reliability,strong feasibility,which can meet the needs of clinical upper limb rehabilitation,and provide a new method for the rehabilitation of upper limb function after stroke.Compared with the traditional upper limb training,it is highly targeted and interesting,and is expected to improve the efficiency of rehabilitation treatment.