基于超声影像分析的功能性电刺激肌肉形态及活动研究

发布时间：2018-07-13 10:54

【摘要】：近年来,脊髓损伤导致瘫痪的发病率呈显著上升趋势。肢体功能重建是截瘫患者康复治疗时关注的一个重点与难点。在肢体功能重建方面,功能性电刺激(functional electrical stimulation, FES)技术被普遍认为是一种有效的临床工具。但是,很多新的FES技术还只局限于实验室阶段,临床应用的FES刺激模式以及达到的效果都非常有限,其原因之一就是对于FES作用的机理方面的研究还有所欠缺,目前还鲜有研究在宏观水平上说明FES如何影响肌肉的功能活动和状态。而骨骼肌的收缩和舒张是人体各种运动的基础,骨骼肌的结构是其功能活动的首要决定因素。所以,要想利用FES实现骨骼肌功能重建并使其得到快速发展和广泛应用的目的,就必然需要对FES作用下骨骼肌的结构及其功能活动与状态的机理进行更深入的探究。本研究设计了等角和等矩两种收缩状态,每种状态包含FES诱发和自主运动两种运动模式。利用B型超声成像系统对FES下的目标肌肉进行成像,提取目标肌肉的功能活动信息—肌肉厚度和纹理特征,同时采集关节角度并将其作为反馈信号,采集相应角度的自主收缩下肌肉的超声图像信息。在每种收缩状态下,本研究提取了FES诱发和自主运动下受试者的肌肉厚度和纹理特征。首先,计算了FES下肌肉厚度、纹理特征与关节角度的相关性,结果显著,表明周期性FES是周期性肌肉活动的直接原因;其次,在达到相同的运动效果前提下,利用比值对比了FES诱发的和自主运动产生的肌肉特征信息的差异,然后根据受试者的个体差异性分析造成这种差异的影响因素,并利用逐步筛选的多元线性回归方法对影响因素进行主因素分析,从而确定每个肌肉特征的主要影响因素,旨在通过调节FES的刺激模式和强度以调整主因素,使得FES诱发下的肌肉特征参数向着达到相同运动效果下的自主收缩的肌肉特征参数回归。本研究利用支持向量机和人工神经网络的机器学习方法,利用FES条件下的既定刺激强度、肌肉厚度和纹理信息对关节角度进行预测。结果表明,采用肌肉信息与FES相结合的方法在预测效果上明显优于只有FES的预测,充分体现了肌肉信息在反映肌肉活动和状态上了重要作用,能为临床FES系统提供了一种潜在的精确控制方法。
[Abstract]:In recent years, the incidence of paralysis caused by spinal cord injury has increased significantly. Limb function reconstruction is an important and difficult point in rehabilitation treatment of paraplegic patients. In limb function reconstruction, functional electrical stimulation (functional electrical stimulation, FES) is widely regarded as an effective clinical tool. However, many new FES technologies are confined to the laboratory stage, and the clinical application of FES stimulation mode and the effect achieved are very limited, one of the reasons is that the mechanism of FES action is still lacking. There are few studies on how FES affects muscle function and state at macro level. The contraction and relaxation of skeletal muscle are the basis of all kinds of exercise, and the structure of skeletal muscle is the primary determinant of its function. Therefore, in order to use FES to reconstruct skeletal muscle function and make it develop rapidly and widely, it is necessary to probe into the structure, the mechanism of functional activity and state of skeletal muscle under the action of FES. In this study, two kinds of contraction states, equal angle and equal moment, are designed. Each state includes two motion modes: FES induced motion and autonomous motion. The target muscles under FES are imaged by B-mode ultrasound imaging system, and the functional activity information of target muscles is extracted-muscle thickness and texture features. At the same time, the angle of joint is collected and used as feedback signal. To collect the ultrasonic image information of muscle under the corresponding angle of autonomic contraction. In each contractive state, we extracted the muscle thickness and texture features of FES-induced and autonomic subjects. Firstly, the correlation between muscle thickness, texture feature and joint angle under FES is calculated. The results show that periodic FES is the direct cause of periodic muscle activity. The ratio was used to compare the differences of FES induced and autonomic muscle characteristic information, and then the influencing factors of the differences were analyzed according to the individual differences of the subjects. The main factors of each muscle feature were analyzed by stepwise multivariate linear regression method. The main factors were adjusted by adjusting the stimulation mode and intensity of FES. The FES induced muscle characteristic parameters are regressed to the muscle characteristic parameters with the same exercise effect. In this study, support vector machine (SVM) and artificial neural network (Ann) were used to predict the angle of joint by using the given stimulation intensity, muscle thickness and texture information under the condition of FES. The results show that the method of combining muscle information with FES is superior to only FES in predicting effect, which fully reflects the important role of muscle information in reflecting muscle activity and state. It can provide a potential accurate control method for clinical FES system.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R310

【参考文献】