User-guided interface for motion tracking in controlled functional electrical stimulation

The need for rehabilitation arises from injuries and diseases that impair movement and independence. Traumas, accidents, and neurological or degenerative disorders- such as stroke, spinal cord, and brain injuries- are major causes of long-term disability, often resulting in partial or total loss of motor control. Stroke, in particular, is among the leading causes of disability worldwide, frequently affecting upper-limb functionality and limiting patients' autonomy. Rehabilitation thus plays a crucial role in restoring motor abilities and promoting recovery through targeted and personalized interventions. In recent decades, alongside the rapid development of technology and computer science, physical rehabilitation has increasingly benefited from advanced tools that support clinical practice. Among these, one of the most relevant is Functional Electrical Stimulation (FES), which uses low-energy electrical pulses to retrain- or, in same case, restore- functional mobility in patients with neuromuscular disorders. The effectiveness of FES can be enhanced by modulating the stimulation delivery according to the natural activation patterns of muscles. In this context, Average Threshold Crossing (ATC) represents an event-driven approach applied to Surface Electromyography (sEMG) signals to estimate muscle contraction force and optimizing FES therapy. A key limitation of this technique is its lack of objectivity. Currently, pulse width and frequency are adjusted manually, forcing clinicians to rely on personal experience to identify suitable parameters. Moreover, therapy effectiveness is often evaluated through subjective observation rather than quantitative metrics, affecting reproducibility and comparability of outcomes. To address these issues, this project aims to develop an innovative system designed to complement an ATC-based FES approach, the Lazarus FES device, providing an objective and patient-specific support tool. By analyzing real-time movements using YOLOv8-pose, the system can suggest to physicians the most appropriate set of stimulation parameters, thereby supporting clinical decision-making, enhancing the effectiveness of FES therapy, and increasing the objectivity of its application. The evaluation process is further refined through the extraction of quantitative features that allow a more accurate and reproducible assessment of patient's motor performance and therapy outcomes. Finally, the project includes the development of an experimental protocol aimed at testing the system first on healthy subjects and, at a later stage, on patients affected by chronic stroke. These steps represent a key phase toward transforming the proposed device into a telemedicine tool. This thesis is structured into seven chapters. The first introduces rehabilitation and provides an overview of stroke as a major cause of motor impairment. The second outlines upper limb assessment techniques and main rehabilitation approaches, both conventional and FES-based. The third describes the Lazarus FES device, the fourth presents the RehabCam software and the experimental protocol, the fifth discusses the results, the sixth outlines the forthcoming clinical trial, and the last chapter summarizes the key findings and conclusions.