Toward a Wearable Gait Lab for Quantitative Assessment of Musculoskeletal Function
Donatella Simonetti is a PhD student in the department Neuromechanical Engineering. Promotors are prof.dr.ir. M. Sartori and prof.dr.ir. H.F.J.M. Koopman from the faculty of Engineering Technology.
Evaluating an individual's musculoskeletal function to understand the mechanical forces exerted by specific muscles and the resulting joint torques is crucial for comprehending movement mechanisms and designing effective training and rehabilitation protocols. In the case of neurologically impaired individuals, such as stroke survivors, the primary goal of rehabilitation is to restore locomotion function, as walking recovery significantly influences quality of life.
In clinical settings, swiftly obtaining quantitative, non-invasive measurements of individual muscle forces poses a challenge. Musculoskeletal assessment often relies on either fast, standardized observational tools for qualitative evaluations of endurance and functional capability or fully equipped laboratories with multiple sensors for quantitative evaluations of musculoskeletal performance at the level of individual joints and muscles. While observational tools are cost-effective and quick, they lack the ability to provide quantitative assessments over time. On the other hand, laboratory-based assessments offer detailed quantitative information but are time-consuming and may not be feasible in all clinical settings. Neither approach simultaneously offers simplicity, rapidity, and quantitative evidence of muscle strength, all of which are crucial in standard clinical rehabilitation.
This work aims to achieve rapid and quantitative measurements of musculoskeletal function through a combination of wearable sensors, advanced musculoskeletal modeling, and signal-processing techniques. Three studies are presented, systematically addressing key aspects in the development of a smart wearable tool for assessing musculoskeletal function.
The first study introduces a fully automated muscle localization algorithm to streamline the identification of muscle sites, reducing the need for manual labor. This algorithm is combined with an electromyography (EMG)-sensorized leg garment and an EMG-driven model to estimate dorsi-plantar flexion ankle torque during dynamic tasks performed by healthy participants. The second study enhances the entire pipeline to generalize across musculoskeletal anatomies and neuromuscular control strategies, involving both healthy participants and post-stroke individuals. This is achieved through a novel EMG-equipped garment and an improved muscle localization algorithm. The final study eliminates the reliance on laboratory-based technologies, such as force plates and camera-based tracking systems, by replacing them with five inertial measurement unit (IMU) sensors. The use of IMU sensors enables a fully wearable technology for the non-invasive estimation of musculoskeletal parameters, including joint angles, muscle activations, and ankle torque resulting from individual muscle forces.
In conclusion, this dissertation demonstrates that wearable and automated technologies offer a viable alternative to standard laboratory techniques. These technologies have the potential to save experimental time, a crucial factor in clinical applications. Additionally, wearable and automated technologies could facilitate the integration of advanced EMG-driven modeling pipelines in clinics, as well as in recreational and occupational settings.
