부산대학교 도서관

The application of short trains of electric pulses, of appropriate frequency and power, to muscle or nerve tissue can induce action potentials which in turn lead to muscle contractions. Timing these action potentials appropriately can help enhance functional movements, a methodology called Functional Electrical Stimulation (FES). The motor and sensory recruitment of muscle and nerves by means of FES has been demonstrated to elicit short- and long-term neurophysiological changes in the central nervous system and thus constitutes a powerful therapeutic rehabilitation tool. The most significant signal parameters affecting the quality of muscle contraction and duration of cortical reorganization caused by Functional Electrical Stimulation Therapy (FEST) are pulse intensity (amplitude), frequency, pulse width and the time delay between pulses. While these parameters are particular to the stimulating signal waveform, there are other key factors at play which determine both the duration and quality of changes. Such factors are electrode placement and synchronization of voluntary command with the induced contraction, which partially determine the successful completion of a movement task. The work presented in this paper aims to study, calibrate, automate and optimize the adaptation process of the aforementioned FEST parameters with respect to induced muscle contraction intensity and perceived discomfort reported by the subject, using both conventional AI and machine learning methodologies.