Computational And Neural Modeling Of Motor Control In Normal Human And Parkinson’s Disease

Study motor control of physiological and pathological movements, is essential for understanding the structure and physiological function of central neural circuits in cerebrum and spinal cord, but also build up the theoretical foundation for motor function rehabilitation and substitution. The two major approaches of studying neural motor control are experimental observation and computational modeling. The experimental measurements of sensorimotor system provide data for developing computational models; and computational simulations in turn provide insights that cannot be obtain from experimental preparation and observation. Therefore, these two approaches could complement and support each other in understanding sensorimotor functions of the central nervous system. The scientific issue that investigated in this study is, the functional modules of central circuits in sensorimotor integration, motor task execution and control, under the dual-control framework. The major breakthrough of this study is to advance the current method and integrate the knowledge of sensorimotor subsystems accumulated in previous studies, and to reveal the physiological and pathological neural control mechanisms from the systematic perspective of view. The major content of this dissertation includes: developing and validating a multi-scale Virtual Arm model, and applying it to investigate the non-linear encoding of joint information in fusimotor control; buiding up an experimental human upper extremity motion collection and analysis platform, combined with modeling approach to reveal the central mechanisms of Parkinsonian tremor, uncovering the correlation between intermuscular synchronization and tremor amplitude, and to provide data for further simulating the pathological mechanism of Parkinsonian tremor using the Virtual Arm model.The major contents of this dissertation are listed as follows:1. Develop the multi-scale integrated Virtual Arm(VA) model based on previous works.The realistic VA model is integrated from previously developed model components of musculoskeletal biomechanics, virtual muscle, muscle spindle, Golgi tendon organ, and newly developed spinal reflex circuits and propriospinal neuronal network. All the components are transplanted into SIMULINK platform with optimized parameters and unified interfaces. The VA model is build up with a modularized structure, which embodies features of easy substitution of customized model components and strong expandability. This model shows great potential in simulating normal and pathological neural motor behaviors, and studying neural motor control mechanisms, as well as in providing a reliable and efficient environment for rehabilitation application development and preclinical evaluation.2. Validate the authenticity of Virtual Arm model by neuro-mechanical behaviors, and evaluate the contribution of feedforward and feedback control to endpoint stability in multijoint arm control.Hand stiffness evaluates the impedance property of upper extremity, while hand variability describes stable range of hand under intrinsic neural noise. These two estimates of neuromechanical behavior can be used to quantify posture stability. The VA model reproduced neuromechanical behaviors(stiffness and variability) that are consistent with experimental data, which validate the authenticity and effectiveness of the VA model. The current experimental technique could not reveal the systematic contribution of proprioceptive feedback in postural control, leading to existing controversy in explanation of the functional roles of spinal reflex circuits. The VA model enables us to design open-loop and closed-loop simulations to quantify the contribution of proprioception to system impedance and stability in multijoint posture control for the first time. The results show that proprioceptive feedback increase hand stiffness ellipses by 35.75±16.99%(mean±standard deviation) and reduce hand variability ellipses by 49.41±21.19%, which reveal the significant contribution of spinal reflex circuits in resisting internal and external perturbations.3. Study the fusimotor control of spindle sensitivity in central and peripheral coding of joint angles.Experimental observations revealed intimate relation between fusimotor neuronal firings and joint trajectories. This implies that joint angle is encoded in fusimotor commands as a central reference in posture control. The fusimotor commands modulate spindle sensitivity by adjusting intrafusal muscle fiber tension under different joint configuration, result in a linear coding of joint angle in primary afferents. However, how fusimotor commands encode joint information to modulate spindle sensitivity is not clear, due to lack of electrophysiological technique in recording fusimotor neuronal activities during normal movement. We addressed this issue using a computational approach by testing three hypotheses of joint coding modalities in fusimotor commands. The hypotheses are rejected/accepted by validate the consistency between simulated and recorded primary afferent firing rates. The results suggest that the joint angle may be encoded by a nonlinear strategy in fusimotor control signal, which modulate spindle sensitivity to assure robust linear coding of joint angle in primary afferents.4. Study the contribution of inter-muscular synchronization in modulating Parkinsonian tremor intensity, and the neural bases of tremor generation.An experimental method is established to collect joint trajectory and electromyography data. Coherence and cross-correlation analyses are used to investigate rhythmic activities of arm muscles during Parkinsonian tremor. Paired coherence and pool-averaged coherence areproposed to evaluate the degree of inter-muscular synchronization, and its relation to tremor intensity in joints. Analyzing results show significant variations of inter-muscular synchronization in different subjects, which is positively correlated to tremor amplitudes in joints. This result suggests synchronization of involuntary activities in different muscles is a main factor that drive Parkinsonian tremor. This study further implies that in the latest “dimmer-switch” model, the central oscillations originate from basal ganglia and cerebello-thalamo-cortical circuit are processed and integrated at the propriospinal neuronal network into alternating activation patterns in antagonistic muscles, then drive the tremor genesis in joints.In summary, we have developed a research platform based on computational modeling and experimental motion analysis to study the functional roles of spinal neural network in motor control under both normal and pathological conditions. This work on one side provides new methods and insights for further motor control investigations, on the other side provides theoretical and methodological bases for rehabilitation application design and preclinical evaluation.