Klinische Neurophysiologie 2004; 35 - 3
DOI: 10.1055/s-2004-831915

Non-Linear Closed Loop Control and System Identification to Induce Coordinated Finger Movements with Repetitive Peripheral Magnetic Stimulation (RPMS)

B Angerer 1, D Schröder 2, A Struppler 3
  • 1München
  • 2München
  • 3München

The therapeutic effect of the RPMS can be increased by a closed loop control to induce coordinated movements in the forearm and the (index) finger since these movements elicit a quite physiological proprioceptive inflow to the CNS. To induce coordinated movements the main task is to develop a closed loop control for a single movement. In the view of control technology the control loop is non-linear and time varying and cannot be treated with standard (linear) control methods. Hence the capabilities of system identification are used to take non-linear characteristics into account, while the RPMS is applied. Such characteristics are the position and orientation of the stimulation coil, the concomitant muscle fatigue, the spasticity, the remaining voluntary activity and the recovery. The system identification achieves an appropriate model of muscle contractions induced by RPMS by minimizing the difference between the measured contraction and the modeled contraction. The recruitment characteristic and dynamics (results of the system identification) are used to compensate the non-linear recruitment due to RPMS. Together with a proportional (for adequate dynamics) plus integral (to guarantee steady state accuracy due to spasticity and remaining voluntary activity) controller, a closed loop control to induce a single movement is achieved. The PI controller calculates such a stimulation intensity, so that the desired position (of the index finger) is reached. To enhance the closed loop control, the stimulation frequency is also adapted according to the variation of the PI controller output. This leads to a more efficient stimulation and a much more physiological proprioceptive inflow to the CNS. To induce coordinated movements the closed loop control is implemented for each single movement. In combination with desired trajectories (measurement results form healthy subjects) this leads to coordinated movements induced by RPMS. As will be presented, the accuracy of the system identification and therefore the quality of the closed loop control depends highly on the remaining voluntary activity. Hence, it is essential to take the voluntary activity into account. For this purpose an RPMS-artifact suppressing EMG amplifier has to be developed. With EMG and the system identification, the voluntary and the induced activity can be distinguished. Hence it is possible to quantify the therapeutic outcome while the RPMS is applied.