Towards closed-loop system for FES control

Stroke is the second leading cause of death in Europe and often results in severe impairment. For individuals with Spinal Cord Injury (SCI), therapeutic techniques help in restoring lost motor functions, preventing muscle atrophy, improving blood circulation, managing pain, and enhancing autonomy. The rehabilitative treatment of the Functional Electrical Stimulation (FES) uses electrical pulses to stimulate muscles at the level of the neuromuscular junctions to recover their physiological activity. Since the efficacy of FES relies on the low-energy pulses injected into the patient, it is essential to monitor the patient’s muscle activity during the FES to regulate stimulation energy. An open challenge in research seeks to automate the adjustment of stimulation patterns based on the subject’s response within a closed-loop system. The aim of this thesis project is to implement a closed-loop FES control system based on a biofeedback signal that provides details on muscle contractions and the movement they induce. The selected biofeedback signals included MechanoMyoGram (MMG), which monitors muscle activity by detecting mechanical vibrations during contraction, and Euler’s angles to track articular movement. The wearable devices developed by the eLiONS laboratory team have been used to collect these signals. These devices are embedded with an Inertial Measurement Unit (IMU), which records acceleration data from which MMG signals can be extracted, and a gyroscope that records angular velocity from which Euler’s angles are obtained. The goal of the closed-loop system is to dynamically adjust the stimulation input in real-time based on feedback from the joint angle and muscle activity. The muscle studied in this project is the biceps brachii during elbow flexion. The synchronized operations between stimulation patterns and biofeedback signals were achieved by developing a real-time application using Object-Oriented Programming (OOP) in Python. This included creating a closed-loop system that effectively integrates and utilizes the combined data. The developed closed-loop system drives the FES by adapting the stimulation pattern based on two factors: the muscle contractions detected from the MMG signal and the comparison between the angle achieved during movement and the subject’s Active Range Of Motion (AROM). These parameters give two feedbacks that together produce an output related to the stimulation’s effectiveness: a negative output causes the increase of the stimulation energy, whereas a positive output shows that the optimal stimulation energy has been identified. Moreover, ten healthy volunteers participated in a simulated clinical protocol focused on investigating the system’s capability to distinguish between a purely stimulated contraction and one where the voluntary contribution from the subjects is present. This test revealed that the closed-loop system could also precisely detect voluntary contributions with an accuracy of 90 % when performing FES contraction with a weight corresponding to 30 % of the Maximal Voluntary Contraction (MVC). Furthermore, a study involving two participants showed that an elastic band opposed to the movement, simulating a sick patient, detected with 87 % accuracy the voluntary contribution. Even though these preliminary findings are limited to a single muscle and movement, there is an opportunity for improvement. Nevertheless, this project demonstrated the potential for a closed-loop FES system driven by examining MMG and Euler’s angle signals.