功能性电刺激助行中的动力和运动学双模态控制信号研究
|
摘要
近年来,脊髓损伤的发病率呈显著上升的趋势,运用功能性电刺激(FES)技术进行脊髓损伤治疗和功能重建研究正成为生物信息学、康复医学及神经电生理等领域的前沿热点课题,其关键技术之一是FES控制信号的选择。现有的FES控制信号只能被动地由外界给出,其康复模式需要长期训练,无法根据动作状态实时加以反馈调节,因而具有自适应性差、易受干扰、难以学习掌握等致命弱点,制约了FES技术的功能重建效果,成为其在康复临床推广与应用中亟待解决的瓶颈问题。
本文从人体动力学和运动学信息出发,提出了一种全新的动力学与运动学双模态FES自动控制方案,并通过对两类控制信号的识别分析验证了该控制方案的可行性。研究中首先设计了一套以单片机微处理器为核心的下肢助行FES系统,包括控制电路、刺激电路和人机界面,通过相关临床实验案例初步验证了该系统对于脊髓损伤的康复治疗和功能重建具有良好效果。用于动力学模态的系统控制信号是来自FES助行过程中步行器测力系统监测到的柄反作用矢量(HRV)信息,文中利用模糊C均值和K均值聚类方法从HRV中提取下肢动作侧向性特征,并对这些特征进行基于交叉验证与支持向量机的分类识别,结果表明在区分左右下肢侧向性信息方面,单个个体的分类正确率最高可达96%,10例实验对象混合分类正确率可达86%;用于运动学模态的系统控制信号是来自FES刺激过程中的膝关节角度信息,文中通过构造适用于下肢FES动作的非线性自回归肌肉模型,设计相关刺激动作实验,建立起了膝关节角度与刺激强度之间的关系,得到该肌肉数学模型的结构与参数,并重点研究了人体运动学信息作为反馈控制信号的神经网络整定PID系统的控制结构和算法,相关实验表明反馈信号与预设角度值之间的误差可控制在5%以内。
本文研究结果表明,基于柄反作用矢量与膝关节角度的动力学与运动学双模态信息有望作为肢体刺激侧向性触发与反馈调整的有效信号,用于助行FES系统的自动控制,从而为更先进的下肢人工运动神经假体系统设计提供帮助。
In recent years, the incidence of spinal cord injury (SCI) disease reveals a general trend of fast increase. It has been a hot topic in bioinformatics, rehabilitation medicine and electro-neurophysiology using functional electrical stimulation (FES) for SCI therapy and function restoration. One of FES key techniques is its control signals choosing. Existing FES control signals are usually given by outside environment passively. The therapy modes need long term training and cannot present feedback to modify the stimulation level according to movement conditions, which leads to many disadvantages, such as poor self-adaptability, great disturbance and hard exercitation, and constrains the function restoration effect of FES. This problem has been the bottleneck to the popularization and application in clinics and should be solved urgently.
From the viewpoint of human kinetics and kinematics information, the thesis presented a new automatic FES control method based on dual-modal of kinetics and kinematics and tested its feasibility through the recognition and analysis for both control signals. A FES system of lower limbs for assisted walking was designed based on chip microprocessor, which included control circuit, stimulation circuit and human machine interface. The effects of this system on SCI rehabilitation therapy and function restoration were preliminarily tested through clinic trails. The system control signal for kinetics mode was the handle reaction vector (HRV) collected from a walker dynamometer system during FES-assisted walking. The lateral features of lower limbs movement were extracted from HRV through fuzzy C-means and K-means clustering analytical method and recognized by K-fold cross-validation and support vector machine. The results showed that the highest recognition rate reached 96% for individual and 86% for group of 10 subjects. The system control signal for kinematics mode was the knee joint angle during FES-action. A non-linear autoregressive muscle model was built for FES action of lower limbs and its structure and parameters were determined by relating the stimulation intensity and knee joint angle in experiments. The research emphasis was laid on the control structure and algorithm of PID system modulated by artificial neural network based on feedback control signal of human kinematics information. Results showed that the relative error between the preestablished angle and the feedback signal was less than 5%.
This research implied that the dual-modal information of kinetics and kinematics based on HRV and knee joint angle is quite hopeful to be the lateral stimulation trigger and feedback signal for the FES automatic control for assisted-walking and give helps to the advanced motor neuroprosthesis system design for lower limbs.
本文从人体动力学和运动学信息出发,提出了一种全新的动力学与运动学双模态FES自动控制方案,并通过对两类控制信号的识别分析验证了该控制方案的可行性。研究中首先设计了一套以单片机微处理器为核心的下肢助行FES系统,包括控制电路、刺激电路和人机界面,通过相关临床实验案例初步验证了该系统对于脊髓损伤的康复治疗和功能重建具有良好效果。用于动力学模态的系统控制信号是来自FES助行过程中步行器测力系统监测到的柄反作用矢量(HRV)信息,文中利用模糊C均值和K均值聚类方法从HRV中提取下肢动作侧向性特征,并对这些特征进行基于交叉验证与支持向量机的分类识别,结果表明在区分左右下肢侧向性信息方面,单个个体的分类正确率最高可达96%,10例实验对象混合分类正确率可达86%;用于运动学模态的系统控制信号是来自FES刺激过程中的膝关节角度信息,文中通过构造适用于下肢FES动作的非线性自回归肌肉模型,设计相关刺激动作实验,建立起了膝关节角度与刺激强度之间的关系,得到该肌肉数学模型的结构与参数,并重点研究了人体运动学信息作为反馈控制信号的神经网络整定PID系统的控制结构和算法,相关实验表明反馈信号与预设角度值之间的误差可控制在5%以内。
本文研究结果表明,基于柄反作用矢量与膝关节角度的动力学与运动学双模态信息有望作为肢体刺激侧向性触发与反馈调整的有效信号,用于助行FES系统的自动控制,从而为更先进的下肢人工运动神经假体系统设计提供帮助。
In recent years, the incidence of spinal cord injury (SCI) disease reveals a general trend of fast increase. It has been a hot topic in bioinformatics, rehabilitation medicine and electro-neurophysiology using functional electrical stimulation (FES) for SCI therapy and function restoration. One of FES key techniques is its control signals choosing. Existing FES control signals are usually given by outside environment passively. The therapy modes need long term training and cannot present feedback to modify the stimulation level according to movement conditions, which leads to many disadvantages, such as poor self-adaptability, great disturbance and hard exercitation, and constrains the function restoration effect of FES. This problem has been the bottleneck to the popularization and application in clinics and should be solved urgently.
From the viewpoint of human kinetics and kinematics information, the thesis presented a new automatic FES control method based on dual-modal of kinetics and kinematics and tested its feasibility through the recognition and analysis for both control signals. A FES system of lower limbs for assisted walking was designed based on chip microprocessor, which included control circuit, stimulation circuit and human machine interface. The effects of this system on SCI rehabilitation therapy and function restoration were preliminarily tested through clinic trails. The system control signal for kinetics mode was the handle reaction vector (HRV) collected from a walker dynamometer system during FES-assisted walking. The lateral features of lower limbs movement were extracted from HRV through fuzzy C-means and K-means clustering analytical method and recognized by K-fold cross-validation and support vector machine. The results showed that the highest recognition rate reached 96% for individual and 86% for group of 10 subjects. The system control signal for kinematics mode was the knee joint angle during FES-action. A non-linear autoregressive muscle model was built for FES action of lower limbs and its structure and parameters were determined by relating the stimulation intensity and knee joint angle in experiments. The research emphasis was laid on the control structure and algorithm of PID system modulated by artificial neural network based on feedback control signal of human kinematics information. Results showed that the relative error between the preestablished angle and the feedback signal was less than 5%.
This research implied that the dual-modal information of kinetics and kinematics based on HRV and knee joint angle is quite hopeful to be the lateral stimulation trigger and feedback signal for the FES automatic control for assisted-walking and give helps to the advanced motor neuroprosthesis system design for lower limbs.
引文
[1]高秀来,人体解剖学,北京:北京大学医学出版社,2004,229~278
[2]张光铂,我国脊柱脊髓损伤基础研究、临床、康复的现状与展望,中国康复医学杂志,2002,17(4): 201~202
[3] Facts and figures at a glance, The National SCI Statistical Center, 2008, 6
[4]胥少汀,脊髓损伤的基础与临床,北京:人民卫生出版社,1993,281~363
[5]李新志,脊髓损伤治疗的研究现状和进展,中国伤残医学杂志,2007,15(2): 94~96
[6]王善金,脊髓损伤药物治疗实验研究进展,国际骨科学杂志,2007,28(6): 384~386
[7]蓝宁,功能性电刺激的原理、设计与应用(一),中国康复理论与实践,1997,3(4): 151~155
[8] Liberson WT, Functional electrotherapy in stimulation of the personal nerve synchronized with the swing phase of th gait of hemiplegic patients, Archives of Physical Medicine and Rehabilitation, 1961, 42: 101~107
[9] Rushton DN, Topical review functional electrical stimulation. Physiological Measurement, 1997(18): 241~275
[10]Graupe D, Functional electrical stimulatior for ambulation by paraplegics: twelve years of clinical observations and system studies, Malabar: Krieger Publishing Company, 1994
[11]Graupe D, Microprocessor-controlled enhanced multiplexed functional electrical stimulator for surface stimulation in paralyzed patients,美国专利,5081989, 1992-05-21
[12]Milos R, Transcutaneous electrical stimulation technology for functional electrical trerapy applications, Proceedings of the 28th IEEE EMBS annual international conference, 2006: 2142~2145
[13]Mirjana P, Simplified arm control in goal-directed movement , Proceeding of the Twelfth Southern Biomedical Engineering Conference , 1993: 31~33
[14]Charles C, Electrical stimulation to restore vestibular function–development of a 3-D vestibular prosthesis, Proceedings of the 2005 IEEE Engineering in Medicine and Biology 27th Annual Conference, Shanghai, China, 2005: 7380~7385
[15]Previdi F, Closed-loop control of FES supported standing up and sitting down using virtual reference feedback tuning, Control Engineering Practice, 2005(13): 1173~1182
[16] Knaflitz M, Suppression of stimulation artifacts from myoelectric-evoked potential recordings, IEEE Trans Biomed Eng 1988, 35(9): 758~763
[17]Popovic MR, Surface stimulation technology for grasping and walking neuroprostheses, IEEE Eng Med Biol Magazine, EMBS, 2001, 20: 82~93
[18]Vossius G, Multichannel stimulation of the lower extremities with surface electrodes, Advances in external control of human extremities IX.ETAN, 1987: 193~203
[19]Haugland MK, Cutaneous whole nerve recording used for correction of footdrop in hemiplegic man, IEEE Transactions on Rehabilitation Engineering, 1995, 3: 307~317
[20]张毓笠,肌电刺激原理及TH型肌力训练与康复仪,医用生物力学,1992,17(4):216~224
[21]王磊,偏瘫患者的健侧对患侧肢体运动的双路肌肉闭环控制,中国康复医学杂志,1990,5(4):164~167
[22]靳静娜,基于FES的下肢运动神经假体系统设计与研究:[硕士学位论文],天津,天津大学,2008
[23]Graupe D, Artificial neural network control of FES in paraplegics for patient responsive ambulation, IEEE Transactions on Biomedical Engineering, 1995, 42(7): 699~707
[24]Rushton DN, Functional electrical stimulation and rehabilitation—an hypothesis, Medical Engineering & Physics, 2003(25): 75~78
[25]蓝宁,功能性电刺激的原理、设计与应用(二),中国康复理论与实践,1998, 4(1):7～13
[26]Daniel Graupe, Functional neuromuscular stimulatior for short-distance ambulation by certain thoracic-level spina-cord-injured paraplegics, Surgical Neurology, 1998, 50: 202~207
[27]毕胜,接受下肢功能性电刺激患者的选择,中国康复医学杂志,2000,15(3):187~188
[28]Donaldson N, A study of handle reaction vector (HRV) in walker FES standing. Proceedings of the Institution of Mechanical Engineers. 1996, 211(1): 81～94
[29]Apostol TM. Linear algebra: A first course, with applications to differential equations. New York: John Wiley & Sons, 1997, 221～251
[30]Babic J, Karcnik T, Bajd T. Stability analysis of four-point walking. Gait Posture. 2001, 14(1): 56～60
[31]Bezdek JC, Pattem recognition with fuzzy objective function algorithms, New York, Plenum Press, 1981
[32]张敏,基于划分的模糊聚类方法.软件学报,2004,15(6):858~868.
[33]刘蕊洁,模糊C聚类均值算法,重庆工学院学报,2008,22(2):139~141
[34]孙晓霞,模糊C-均值(FCM)聚类算法的实现,计算机应用于软件,2008,25(3):48~50
[35]Wilpon JG, Modified K-Means clustering algorithms for use in isolated word recognition. IEEE Transactions on Acoustics,Speech,Signal Proc,1985,33(3):587~594
[36]王红睿,均衡化的改进K均值聚类法,吉林大学学报,2006,24(2):172~176
[37]Ding CHQ, Multi-class protein fold recognition using support vector machines and neural networks. Bioinformatics, 2001, 17(4): 349～358
[38]Burges CA, Tutoring on support vector machines for pattern recognition. Data Mining and Knowledge Discovery, 1998,2(2):121~167
[39]张学工,统计学习理论的本质,北京:清华大学出版社,2000
[40]尚禹,电刺激骨骼肌产生收缩肌力的数学模型,国际生物医学工程杂志,2007,30(2):77~80
[41]Previdi F, Identification of a class of non-linear parametrically varying models, International Journal of Adaptive Control and Signal Processing, 2003, 17: 33~50
[42]Chen S, Representations of Non-linear Systems: the NARMAX Model, 1989, 49: 1013~1032
[43]时文飞,基于人工神经网络与PID的复合控制研究:[硕士学位论文],重庆,重庆大学,2007
[44]刘金琨,先进PID控制MATLAB仿真(第2版),北京,电子工业出版社,162~165
[2]张光铂,我国脊柱脊髓损伤基础研究、临床、康复的现状与展望,中国康复医学杂志,2002,17(4): 201~202
[3] Facts and figures at a glance, The National SCI Statistical Center, 2008, 6
[4]胥少汀,脊髓损伤的基础与临床,北京:人民卫生出版社,1993,281~363
[5]李新志,脊髓损伤治疗的研究现状和进展,中国伤残医学杂志,2007,15(2): 94~96
[6]王善金,脊髓损伤药物治疗实验研究进展,国际骨科学杂志,2007,28(6): 384~386
[7]蓝宁,功能性电刺激的原理、设计与应用(一),中国康复理论与实践,1997,3(4): 151~155
[8] Liberson WT, Functional electrotherapy in stimulation of the personal nerve synchronized with the swing phase of th gait of hemiplegic patients, Archives of Physical Medicine and Rehabilitation, 1961, 42: 101~107
[9] Rushton DN, Topical review functional electrical stimulation. Physiological Measurement, 1997(18): 241~275
[10]Graupe D, Functional electrical stimulatior for ambulation by paraplegics: twelve years of clinical observations and system studies, Malabar: Krieger Publishing Company, 1994
[11]Graupe D, Microprocessor-controlled enhanced multiplexed functional electrical stimulator for surface stimulation in paralyzed patients,美国专利,5081989, 1992-05-21
[12]Milos R, Transcutaneous electrical stimulation technology for functional electrical trerapy applications, Proceedings of the 28th IEEE EMBS annual international conference, 2006: 2142~2145
[13]Mirjana P, Simplified arm control in goal-directed movement , Proceeding of the Twelfth Southern Biomedical Engineering Conference , 1993: 31~33
[14]Charles C, Electrical stimulation to restore vestibular function–development of a 3-D vestibular prosthesis, Proceedings of the 2005 IEEE Engineering in Medicine and Biology 27th Annual Conference, Shanghai, China, 2005: 7380~7385
[15]Previdi F, Closed-loop control of FES supported standing up and sitting down using virtual reference feedback tuning, Control Engineering Practice, 2005(13): 1173~1182
[16] Knaflitz M, Suppression of stimulation artifacts from myoelectric-evoked potential recordings, IEEE Trans Biomed Eng 1988, 35(9): 758~763
[17]Popovic MR, Surface stimulation technology for grasping and walking neuroprostheses, IEEE Eng Med Biol Magazine, EMBS, 2001, 20: 82~93
[18]Vossius G, Multichannel stimulation of the lower extremities with surface electrodes, Advances in external control of human extremities IX.ETAN, 1987: 193~203
[19]Haugland MK, Cutaneous whole nerve recording used for correction of footdrop in hemiplegic man, IEEE Transactions on Rehabilitation Engineering, 1995, 3: 307~317
[20]张毓笠,肌电刺激原理及TH型肌力训练与康复仪,医用生物力学,1992,17(4):216~224
[21]王磊,偏瘫患者的健侧对患侧肢体运动的双路肌肉闭环控制,中国康复医学杂志,1990,5(4):164~167
[22]靳静娜,基于FES的下肢运动神经假体系统设计与研究:[硕士学位论文],天津,天津大学,2008
[23]Graupe D, Artificial neural network control of FES in paraplegics for patient responsive ambulation, IEEE Transactions on Biomedical Engineering, 1995, 42(7): 699~707
[24]Rushton DN, Functional electrical stimulation and rehabilitation—an hypothesis, Medical Engineering & Physics, 2003(25): 75~78
[25]蓝宁,功能性电刺激的原理、设计与应用(二),中国康复理论与实践,1998, 4(1):7～13
[26]Daniel Graupe, Functional neuromuscular stimulatior for short-distance ambulation by certain thoracic-level spina-cord-injured paraplegics, Surgical Neurology, 1998, 50: 202~207
[27]毕胜,接受下肢功能性电刺激患者的选择,中国康复医学杂志,2000,15(3):187~188
[28]Donaldson N, A study of handle reaction vector (HRV) in walker FES standing. Proceedings of the Institution of Mechanical Engineers. 1996, 211(1): 81～94
[29]Apostol TM. Linear algebra: A first course, with applications to differential equations. New York: John Wiley & Sons, 1997, 221～251
[30]Babic J, Karcnik T, Bajd T. Stability analysis of four-point walking. Gait Posture. 2001, 14(1): 56～60
[31]Bezdek JC, Pattem recognition with fuzzy objective function algorithms, New York, Plenum Press, 1981
[32]张敏,基于划分的模糊聚类方法.软件学报,2004,15(6):858~868.
[33]刘蕊洁,模糊C聚类均值算法,重庆工学院学报,2008,22(2):139~141
[34]孙晓霞,模糊C-均值(FCM)聚类算法的实现,计算机应用于软件,2008,25(3):48~50
[35]Wilpon JG, Modified K-Means clustering algorithms for use in isolated word recognition. IEEE Transactions on Acoustics,Speech,Signal Proc,1985,33(3):587~594
[36]王红睿,均衡化的改进K均值聚类法,吉林大学学报,2006,24(2):172~176
[37]Ding CHQ, Multi-class protein fold recognition using support vector machines and neural networks. Bioinformatics, 2001, 17(4): 349～358
[38]Burges CA, Tutoring on support vector machines for pattern recognition. Data Mining and Knowledge Discovery, 1998,2(2):121~167
[39]张学工,统计学习理论的本质,北京:清华大学出版社,2000
[40]尚禹,电刺激骨骼肌产生收缩肌力的数学模型,国际生物医学工程杂志,2007,30(2):77~80
[41]Previdi F, Identification of a class of non-linear parametrically varying models, International Journal of Adaptive Control and Signal Processing, 2003, 17: 33~50
[42]Chen S, Representations of Non-linear Systems: the NARMAX Model, 1989, 49: 1013~1032
[43]时文飞,基于人工神经网络与PID的复合控制研究:[硕士学位论文],重庆,重庆大学,2007
[44]刘金琨,先进PID控制MATLAB仿真(第2版),北京,电子工业出版社,162~165