The work presented in this manuscript focuses on bridging the gap between emerging Motion Capture (MoCap) technologies for telerehabilitation, particularly Inertial Measurement Units (IMUs), and current shoulder rehabilitation paradigms, which emphasize functional and ecological perspectives on upper-limb recovery. Telerehabilitation has gained substantial attention in recent years due to its potential to reduce hospitalization costs and simplify clinical logistics. This progress has been enabled by technological advancements in MoCap systems. In particular, IMUs represent a promising class of wearable motion capture sensors, as they can estimate the Three-dimensional (3D) orientation of body segments without requiring external infrastructure such as cameras or transmitters. Among the various body regions, the shoulder complex and upper limb pose some of the greatest challenges in physical rehabilitation due to the wide range and complexity of their movements, as well as their fundamental role in interacting with the external environment. Consequently, contemporary upper-limb rehabilitation approaches aim not only to restore the ability to perform daily gestures using pre-injury kinematic patterns (functional paradigm), but also to recover smooth, natural, and unrestricted interaction with the environment (ecological paradigm). Despite this, most IMU-based studies on upper-limb assessment and rehabilitation focus on simple, isolated movements rather than on complex gestures involved in Activities of daily living (ADLs). Therefore, reliable assessment of functional upper-limb movements in ecological conditions remains a major challenge for IMU-based telerehabilitation systems. To promote the adoption of functional and ecological perspectives in IMU-based telerehabilitation, this thesis investigates novel and improved IMU data processing techniques and algorithms through multiple complementary research studies and technological development. The work builds upon Sentry, an IMU-based wearable device originally designed as a movement monitoring system for continuous assessment of the shoulder complex. The device was conceptualized by the REHELab research group at the University of Genoa (Savona, Italy) and developed by Swhard s.r.l. (Genoa, Italy). In this research, Sentry was extended to incorporate four IMUs, enabling full upper-limb tracking. A dataset of glenohumeral joint kinematics during simulated ADLs was collected from two cohorts: a young population (25.65 ± 2.60 years) and an elderly population (70.80 ± 6.97 years). The dataset was used to explore different methods for the evaluation and clinical assessment of functional upper-limb gestures. The first study focused on the development and validation of an algorithm to score repeated functional movements performed during rehabilitation by comparing them with a prescribed reference template. The algorithm performs a spatial comparison of kinematic trajectories represented through quaternions, yielding a template agreement score. When tested on the collected dataset, the method demonstrated the ability to discriminate between repetitions of the same movement and different movements (p < 0.01, rrb effect size = 0.97, contrast ratio = 1.7), with only mild performance differences relative to traditional approaches (rrb effect size < 0.5). The second study used the dataset to reconstruct and compare angular kinematic trajectories of functional gestures inspired by ADLs between young and elderly participants. The objective was to build a normative reference dataset from the young population and to assess the ability of IMUs to detect age-related alterations in movement patterns. Statistically significant differences (p < 0.05) were observed in pairwise comparisons between the two groups during specific movement phases across all ADLs, consistent with aging-related phenomena reported in the literature. These findings support the capability of IMUs to detect clinically relevant kinematic alterations. The third study investigated the feasibility of using the Finite Helical Axes (FHA) approach, based on axis–angle rotation representation, to pursue the same objectives as the second study. In collaboration with researchers from Vrije Universiteit Brussel, an optimized FHA algorithm was developed and applied to the IMU data. The results demonstrated that the Mean Angle (MA) metric, describing axis dispersion, could distinguish between movements performed by young and elderly participants (p < 0.05, Cohen’s d >0.2 for most movements). Future work may focus on improving visualization of the three-dimensional trajectories of the helical axes. The fourth study analyzed a pilot dataset collected from five unimpaired adult participants who wore the Sentry device for seven hours on two separate days while performing everyday activities. The aim was to assess the feasibility of detecting differences in functional workspace and movement speed between a mobile and a non-mobile arm, and to identify meaningful metrics for quantifying movement in ecological conditions. The results revealed significant differences in functional workspace between the two arms (p < 0.02 in spatial distribution comparisons, distinct clustering patterns identified through k-means inertia curves), as well as differences in average movement speed (p < 0.02). Finally, prototype versions of the main architectural components of a Sentry-based telerehabilitation system were developed, integrating the methods explored in this research. These included patient and clinician applications featuring communication with the Sentry device, movement feedback through an animated avatar combined with the proposed movement scoring algorithm, and a central database for data storage and exchange. Overall, this work provides methodological, computational, and technological foundations for incorporating functional and ecological perspectives into IMU-based telerehabilitation systems. As a first step toward bridging the gap between IMU-based motion analysis and functional–ecological rehabilitation paradigms, this thesis lays the groundwork for the development of next-generation telerehabilitation systems enabling objective, personalized, and real-world upper-limb recovery. Future directions will focus on consolidating the proposed computational approaches, extending validation to impaired populations, completing system integration, and deploying the platform to real users to evaluate usability and clinical relevance. Moreover, integrating the methodologies presented in this work with modern AI techniques will enable more personalized treatments in the context of remote movement monitoring and rehabilitation.
A Functional and Ecological Perspective on Inertial Measurement Unit–Based Shoulder Telerehabilitation: Methods, Algorithms, and System Implementation
IURATO, MATTEO
2026-05-11
Abstract
The work presented in this manuscript focuses on bridging the gap between emerging Motion Capture (MoCap) technologies for telerehabilitation, particularly Inertial Measurement Units (IMUs), and current shoulder rehabilitation paradigms, which emphasize functional and ecological perspectives on upper-limb recovery. Telerehabilitation has gained substantial attention in recent years due to its potential to reduce hospitalization costs and simplify clinical logistics. This progress has been enabled by technological advancements in MoCap systems. In particular, IMUs represent a promising class of wearable motion capture sensors, as they can estimate the Three-dimensional (3D) orientation of body segments without requiring external infrastructure such as cameras or transmitters. Among the various body regions, the shoulder complex and upper limb pose some of the greatest challenges in physical rehabilitation due to the wide range and complexity of their movements, as well as their fundamental role in interacting with the external environment. Consequently, contemporary upper-limb rehabilitation approaches aim not only to restore the ability to perform daily gestures using pre-injury kinematic patterns (functional paradigm), but also to recover smooth, natural, and unrestricted interaction with the environment (ecological paradigm). Despite this, most IMU-based studies on upper-limb assessment and rehabilitation focus on simple, isolated movements rather than on complex gestures involved in Activities of daily living (ADLs). Therefore, reliable assessment of functional upper-limb movements in ecological conditions remains a major challenge for IMU-based telerehabilitation systems. To promote the adoption of functional and ecological perspectives in IMU-based telerehabilitation, this thesis investigates novel and improved IMU data processing techniques and algorithms through multiple complementary research studies and technological development. The work builds upon Sentry, an IMU-based wearable device originally designed as a movement monitoring system for continuous assessment of the shoulder complex. The device was conceptualized by the REHELab research group at the University of Genoa (Savona, Italy) and developed by Swhard s.r.l. (Genoa, Italy). In this research, Sentry was extended to incorporate four IMUs, enabling full upper-limb tracking. A dataset of glenohumeral joint kinematics during simulated ADLs was collected from two cohorts: a young population (25.65 ± 2.60 years) and an elderly population (70.80 ± 6.97 years). The dataset was used to explore different methods for the evaluation and clinical assessment of functional upper-limb gestures. The first study focused on the development and validation of an algorithm to score repeated functional movements performed during rehabilitation by comparing them with a prescribed reference template. The algorithm performs a spatial comparison of kinematic trajectories represented through quaternions, yielding a template agreement score. When tested on the collected dataset, the method demonstrated the ability to discriminate between repetitions of the same movement and different movements (p < 0.01, rrb effect size = 0.97, contrast ratio = 1.7), with only mild performance differences relative to traditional approaches (rrb effect size < 0.5). The second study used the dataset to reconstruct and compare angular kinematic trajectories of functional gestures inspired by ADLs between young and elderly participants. The objective was to build a normative reference dataset from the young population and to assess the ability of IMUs to detect age-related alterations in movement patterns. Statistically significant differences (p < 0.05) were observed in pairwise comparisons between the two groups during specific movement phases across all ADLs, consistent with aging-related phenomena reported in the literature. These findings support the capability of IMUs to detect clinically relevant kinematic alterations. The third study investigated the feasibility of using the Finite Helical Axes (FHA) approach, based on axis–angle rotation representation, to pursue the same objectives as the second study. In collaboration with researchers from Vrije Universiteit Brussel, an optimized FHA algorithm was developed and applied to the IMU data. The results demonstrated that the Mean Angle (MA) metric, describing axis dispersion, could distinguish between movements performed by young and elderly participants (p < 0.05, Cohen’s d >0.2 for most movements). Future work may focus on improving visualization of the three-dimensional trajectories of the helical axes. The fourth study analyzed a pilot dataset collected from five unimpaired adult participants who wore the Sentry device for seven hours on two separate days while performing everyday activities. The aim was to assess the feasibility of detecting differences in functional workspace and movement speed between a mobile and a non-mobile arm, and to identify meaningful metrics for quantifying movement in ecological conditions. The results revealed significant differences in functional workspace between the two arms (p < 0.02 in spatial distribution comparisons, distinct clustering patterns identified through k-means inertia curves), as well as differences in average movement speed (p < 0.02). Finally, prototype versions of the main architectural components of a Sentry-based telerehabilitation system were developed, integrating the methods explored in this research. These included patient and clinician applications featuring communication with the Sentry device, movement feedback through an animated avatar combined with the proposed movement scoring algorithm, and a central database for data storage and exchange. Overall, this work provides methodological, computational, and technological foundations for incorporating functional and ecological perspectives into IMU-based telerehabilitation systems. As a first step toward bridging the gap between IMU-based motion analysis and functional–ecological rehabilitation paradigms, this thesis lays the groundwork for the development of next-generation telerehabilitation systems enabling objective, personalized, and real-world upper-limb recovery. Future directions will focus on consolidating the proposed computational approaches, extending validation to impaired populations, completing system integration, and deploying the platform to real users to evaluate usability and clinical relevance. Moreover, integrating the methodologies presented in this work with modern AI techniques will enable more personalized treatments in the context of remote movement monitoring and rehabilitation.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
