Résumé
The topic of this thesis was the rehabilitation of movement of paralyzed limbs through functional electrical stimulation (FES). The objective of the project was to explore the possibility of using information from sensory nerve fibers of muscle receptors as feedback of the closed-loop control of FES systems using intrafascicular peripheral nerve electrodes.<br />Acute animal experiments were performed to record afferent muscle spindle responses to passive stretch. The recordings were performed using the new thin-film Longitudinal Intra-Fascicular Electrode (tfLIFE), developed by Dr. Ken Yoshida at Aalborg University in Denmark. A first-order model of muscle spindle response to passive muscle stretch was proposed that manages to capture the non-linear properties of the afferent neural activity. Moreover, estimation of muscle state from the recorded multi-channel ENG provided more robust results compared to using single-channel recordings.<br />For the abovementioned model to be usable in a estimator of muscle state, the rate of change of muscle length during movement must have negligible effect on model parameters. A neural spike detection and classification scheme was developed for the purpose of isolating sensory neural activity of muscle receptors having minimal sensitivity to the velocity of muscle motion. The algorithm was based on the multi-scale continuous wavelet transform using complex wavelets. The detection scheme outperforms the commonly used simple threshold detection, especially with recordings having low SNR. Results of classification of units indicate that the developed classifier is able to isolate activity having linear relationship with muscle length, which is a step towards on-line model-based estimation of muscle length that can be used in a closed-loop FES system with natural sensory feedback.<br />One of the main issues limiting the interpretation of ENG data is the low level of the neural signal compared to the level of noise in the recordings. Our hypothesis was that shielding the implant site would help improve signal-to-noise level. Experimental results from a preliminary study indicate that placing a standard cuff electrode around the tfLIFE active sites increases the level of ENG signal in the recordings.