The Impact of FES on Event-driven sEMG Information: A Preliminary Investigation

Stroke and Spinal Cord Injury (SCI) are the two most common causes of paralysis, affecting thousands of people worldwide. Voluntary movement rehabilitation has been enriched in recent years by continuously incorporating knowledge about the mechanisms underlying motor function recovery. One central concept in neurorehabilitation is neuroplasticity, the ability of the central nervous system to reorganize itself during the acquisition, retention, and consolidation of motor skills. Functional Electric Stimulation (FES) aid motor recovery by delivering electrical current pulses to patients through non-invasive electrodes and specialized equipment. An event-based system was previously designed to control FES in real-time. The system allowed accurate replication of the movement performed by a therapist on a patient subject by automatically modulating the intensity of electrical stimulation. For this purpose, a wearable device based on the Average Threshold Crossing (ATC) technique was used to extract meaningful information from the surface ElectroMyoGraphic (sEMG) signal. Stimulation patterns are consequently derived and transmitted to the patient in real-time via surface electrodes but without providing feedback from the patient to the system. The goal of this thesis project represents a step toward optimizing the previously realized system insert a feedback as a further input to modulate the FES parameters and make the therapy more efficient. The idea is to acquire the Threshold Crossing (TC) signal and evaluate muscle activation during the stimulation session in different circumstances. An experimental protocol was established and tested on the thesis candidate to evaluate the behavior of the Biceps Brachii and Quadriceps Femoris muscles acquiring both in the cases of pure stimulation and in combination with voluntary activity in isometric conditions. While the main FES parameters were varied, the voluntary contribution was standardized by measuring the Maximum Voluntary Contraction (MVC), of which the percentages of 20%, 50% and 80% were selected. From these tests, five clusters were defined depending on the total charge injected by the stimulation pattern. Based on them, two main features, i.e., the events width and the number of events, are extracted from the TC signals to characterize the muscular activation. Five participants were selected to perform a subset of appropriate stimulation pattern combinations to compare the results with the preliminary tests. The results showed an increase in both TC-extracted features, contrarily to the preliminary tests in which an increase up to the intermediate cluster and a decrease in higher clusters was shown. To investigate anomalous behaviors, a study was conducted to observe the effect of stimulation on the sEMG signal. In particular conditions, it emerged that a stimulation artifact overlaps the evoked muscle response (M-wave), preventing the assessment of voluntary muscle activity. The contamination of the sEMG signal due to the stimulation artifacts, highlights that the TC signal extraction technique does not always produce the expected results. A new calibration method was tested to overcome this limitation consisting of calibrating the threshold during stimulation to adapt the ATC technique to the morphological changes of the sEMG under stimulation conditions. New tests were performed to validate this technique, applying the following metrics: MVC level of 40%, stimulus amplitudes less than 12 mA, and frequencies from 10 to 50 Hz.