Delsys Prize 2004 Winning Proposal
Estimation of Proprioceptive Reflex Gains using Surface EMG
Professor Dr. F.C.T van der Helm, Delft University of Technology, Dept. of Mechanical Engineering, Delft, The Netherlands.
Robot manipulators are being used to impose force perturbations to the hand. Hand position, hand force and EMG are being recorded. EMG signals are being pre-processed to estimate the dynamic relation to hand force. By optimizing the perturbation signal and using advanced closed loop system identification algorithms, the position, velocity and force feedback originating from the muscle spindles and Golgi tendon organs can be separated. Results were applied to patiens with neurological disorders (CVA, Parkinson).
Goal of the research program is to assess the importance of the contribution of the proprioceptive reflexes for posture maintainance. A closed-loop feedback model incorporating the intrinsic muscle contributions (stiffness and viscosity) and reflexive contributions (muscle spindle and Golgi tendon organs, GTO) was developed. Principally, identification of the separate pathways can only be done using external perturbations. We used force perturbations by a robot manipulator. Three types of manipulators have been used in our lab: 1 DOF shoulder joint, 1 DOF wrist joint and 2 DOF shoulder and elbow joint. Hand position, hand force and EMG of agonist and antagonist muscle were recorded. EMG signals are indispensable to separate the intrinsic and reflexive contributions to posture maintainance, since the reflexive pathways via the Central Nervous System (CNS) will induce small changes in EMG, whereas the intrinsic contribution leads to high co-contraction levels. To distinguish the small EMG changes in a very noisy EMG recording substantial pre-processing of the EMG signal was required. The EMG signal was rectified, but not low-pass filtered. The first step was to estimate the dynamic relation between EMG to hand force in a separate isometric experiment, resulting in an estimate of the acitvation dynamics. Secondly, EMG was pre-whitened. Thirdly, the perturbation signal was crest-optimized to optimize the signal-to-noise ratio. In an advanced closed-loop system identification scheme, the transfer function between hand position and EMG could be estimated, revealing the contribution of the proprioceptive feedback gains as set in the CNS. Next, the position, velocity and force feedback gains could be estimated.
Typically, the Variance-Accounted-For (VAF) values of the mechanical system (force to position) approached 100%, whereas the VAF for position to EMG were between 50% to 70%, which was considered very good given the noisy character of EMG.
The results showed that healthy subjects instantaneously adapt their feedback gains as a response to altered environmental conditions (damping, mass) and to altered frequency content of the perturbation signal (white band noise, narrow band noise). Experiments with patients with neurogical disorders showed their limitation in modulating the reflexive feedback gains. Patients with Parkinson’s disease showed larger reflexive time-delays and higher position feedback gains w.r.t. controls. CVA patients with spasms showed decreased velocity feedback gains, larger position feedback gains and increased reflexive time-delays. Patients with Complex Regional Pain Syndrome and dystonia showed no negative feedback gains in narrow band noise conditions, indicating a dysfunction of inhibitory pathways in the CNS.
Our conclusion is that there is a whole new venue of experimental methods available to study motor control in healthy subjects and patients with neurological disorders. These experimental methods use force-controlled robot manipulators to impose perturbations to the subjects, and use EMG to distinguish between intrinsic and reflexive muscle properties.
Van der Helm FCT, Schouten AC, De Vlugt E, Brouwn GG (2002). Identification of intrinsic and reflexive components of human arm dynamics during postural control. J. Neurosc. Meth 119, 1-14.
Schouten AC, Van de Beek WJT, Van Hilten JJ, Van deer Helm FCT (2003). Proprioceptive reflexes in patients with Reflex Sympathetic Dystrophy. Exp. Brain Res 151, 1-8.
Schouten AC, De Vlugt E, Van der Helm FCT (2004a). Estimating the reflexive impedance during posture tasks. Submitted.
Schouten AC, De Vlugt E, Van Hilten JJ, Van der Helm FCT (2004b). Quantifying reflexes with mechanical admittance and reflexive impedance. Submitted.