Linear identification of nonlinear wrist neuromechanics in stroke

Abstract

In many stroke patients, a motor cortex lesion alters motor control. Initially, paresis is most prominent, but then over time, joint stiffening and hyperreflexia may occur. How these different disorders develop over time is still unknown due to high system complexity. Secondary changes in the corticospinal tract, peripheral biomechanics and spinal reflexive system, may also occur. This thesis is part of the EXPLICIT-Stroke study (see Chapters 1, 2 and 3), a randomized, controlled trial that researches the effect of early therapy on post stroke recovery of the upper limb. Amongst other measurements, the EXPLICIT-Stroke study investigates post-stroke changes of brain function and corticospinal tract with fMRI and TMS, respectively. The work in this thesis aims to identify post stroke changes in peripheral biomechanics and the spinal reflexes of the wrist: wrist joint neuromechanics. Neuromechanics play an important role in the functioning of a joint. Inputs to the neuromechanical system are: neural input originating from supraspinal regions and externally applied rotation/torque. Neuromechanics therefore represent the translation from supraspinal input to muscle contraction and resultant joint rotation, torque and/or stiffness, and also describe the relationship between external perturbation and joint response. Joint impedance, the dynamic relationship between joint angle and resultant joint torque, was used to investigate joint neuromechanics. Neuromechanics can be split into: dynamics of passive soft tissues, voluntary muscle contraction and reflexive muscle contraction. Knowledge of changes in the underlying properties yields insight into the complex development of movement disorders and can eventually lead to targeted therapy. Measurement of impedance is achieved by external (motorised) angular perturbation of the joint whilst measuring the joint torque response. This is commonly supported by measurement of muscle activation: Electromyography (EMG). Joint neuromechanics are highly nonlinear. Although many nonlinear neuromechanical properties are known from literature, the effects of these nonlinear properties on joint impedance, and thus their functional and clinical relevance, have generally not been quantified. Commonly known examples of nonlinearity are increased resistance against movement in extreme angles of the range of motion and increased joint stiffness with muscle contraction. Due to nonlinearity, linearly observed neuromechanics depend on input, i.e., depend on measurement conditions. In line with the previous examples, joint stiffness depends on muscle contraction and joint angle. Therefore, understanding nonlinearity is essential for interpretation of joint impedance. Linear modelling and system identification methods allow for estimation of neuromechanical parameters. Use of these linear methods restricts measurement to small deviations in joint angle, angular velocity and muscle contraction. As normal movement often includes large deviations in angle, angular velocity and muscle contraction, such measurements do not describe the full range of interest in joint neuromechanics. Furthermore, comparison of subjects requires that they are measured in the same angles, angular velocities and contraction levels, such that observed differences between subjects are due to differences in neuromechanical properties, and not due to nonlinearity. For example, high joint stiffness in Chapter 9, was hypothesized to be caused by co-activation of the antagonistic muscle pair, i.e., the nonlinear system under a different contraction level (active state), and not caused by different peripheral neuromechanical properties.