Tesi etd-11152024-075708
Link copiato negli appunti
Tipo di tesi
Dottorato
Autore
GASPAR PINTO RAMOS CORREIA, CAROLINA
URN
etd-11152024-075708
Titolo
Towards Advanced Body-Machine Interfaces for Neuromotor Rehabilitation
Settore scientifico disciplinare
ING-IND/34
Corso di studi
Istituto di Biorobotica - PhD in Biorobotica - PON
Commissione
Membro Prof. MICERA, SILVESTRO
Presidente Prof. STEFANO MAZZOLENI
Membro Prof.ssa SARA MOCCIA
Presidente Prof. STEFANO MAZZOLENI
Membro Prof.ssa SARA MOCCIA
Parole chiave
- Body-machine interfaces
- Sensor-based interfaces
- Electromyography
- Inertial measurement units
- Motor control
- Neuromotor rehabilitation
- Myoelectric interfaces
- Trunk muscle activation
- Elbow flexors and extensors
- Cervical spine kinematics
- Finite helical axis
- VR-based therapy
- Sensorimotor training
Data inizio appello
21/11/2025;
Disponibilità
parziale
Riassunto analitico
This thesis investigates the design, development, and application of sensor-based body-machine interfaces (BoMIs) to support motor control and rehabilitation in individuals with neuromuscular impairments. BoMIs provide a powerful framework for translating biosignals -- such as muscle activity and body motion -- into control signals, enabling a more natural, intuitive interaction with external devices. While they have been increasingly explored for human assistance and augmentation, their potential to guide rehabilitation, promote motor learning, and inform individualized recovery strategies deserves further investigation.
To address this gap, this thesis examines BoMI applications across three experimental contexts, each targeting a key aspect for motor rehabilitation: (1) the development of multimodal BoMIs to guide coordinated activation of multiple muscles; (2) the use of myoelectric BoMIs to characterize fundamental differences in voluntary motor control following neurological injury; and (3) the deployment of motion-driven BoMIs in clinical rehabilitation to support recovery and reveal mechanisms of neuromuscular adaptation.
The first study addresses the challenge of trunk muscle weakness and impaired postural control -- functions that require coordinated recruitment of multiple muscle groups. Traditional rehabilitation approaches often prescribe movement exercises without ensuring proper underlying muscle activation. This study introduces a novel BoMI that combines electromyography (EMG) and motion sensing with machine learning to generate movement trajectories aligned with target muscle activation patterns. When tested in neurotypical individuals, the interface enabled users to engage trunk muscles selectively and consistently, suggesting its potential for future use in postural training among clinical populations.
The second study focuses on the upper limb, examining differences in voluntary control between the biceps and triceps brachii -- antagonistic muscles essential for elbow control and differentially affected by cervical spinal cord injury (SCI). Using a myoelectric interface with real-time EMG feedback, participants tracked dynamic force targets through isometric EMG contractions of each muscle. The triceps exhibited significantly higher EMG control accuracy and lower variability than the biceps, indicating superior precision in modulating voluntary activation. These differences likely reflect asymmetries in descending corticospinal drive, inhibitory regulation, and motor unit recruitment strategies, which may contribute to the greater susceptibility of elbow extensors to functional impairment following SCI.
The third study compares two rehabilitation approaches for persistent neck pain: a motion-driven BoMI with virtual reality (VR) feedback versus conventional therapist-guided training. Participants in both groups completed six weeks of sensorimotor training, with the VR group using motion tracking and real-time visual feedback to guide cervical movements during gamified tasks. While both groups showed improvements in pain and disability, the BoMI-guided VR group exhibited greater enhancements in cervical motion quality and coordination, as quantified through kinematic and finite helical axis analyses. These findings underscore the added value of BoMI-based VR rehabilitation in promoting functional recovery and eliciting neuromechanical adaptations not observed with conventional therapy.
Overall, these studies demonstrate that sensor-based BoMIs can be effectively applied to address key challenges in motor rehabilitation, from guiding selective muscle activation and assessing neuromuscular control to supporting recovery in clinical populations. Across distinct contexts, BoMIs enabled real-time, task-specific interaction driven by biosignals, allowing both functional engagement and deeper analysis of motor performance. By integrating biosignal-based control with physiologically informed task design, this thesis contributes to advancing the use of BoMIs as practical tools for rehabilitation and as experimental platforms for investigating motor control in both neurotypical and clinical populations.
To address this gap, this thesis examines BoMI applications across three experimental contexts, each targeting a key aspect for motor rehabilitation: (1) the development of multimodal BoMIs to guide coordinated activation of multiple muscles; (2) the use of myoelectric BoMIs to characterize fundamental differences in voluntary motor control following neurological injury; and (3) the deployment of motion-driven BoMIs in clinical rehabilitation to support recovery and reveal mechanisms of neuromuscular adaptation.
The first study addresses the challenge of trunk muscle weakness and impaired postural control -- functions that require coordinated recruitment of multiple muscle groups. Traditional rehabilitation approaches often prescribe movement exercises without ensuring proper underlying muscle activation. This study introduces a novel BoMI that combines electromyography (EMG) and motion sensing with machine learning to generate movement trajectories aligned with target muscle activation patterns. When tested in neurotypical individuals, the interface enabled users to engage trunk muscles selectively and consistently, suggesting its potential for future use in postural training among clinical populations.
The second study focuses on the upper limb, examining differences in voluntary control between the biceps and triceps brachii -- antagonistic muscles essential for elbow control and differentially affected by cervical spinal cord injury (SCI). Using a myoelectric interface with real-time EMG feedback, participants tracked dynamic force targets through isometric EMG contractions of each muscle. The triceps exhibited significantly higher EMG control accuracy and lower variability than the biceps, indicating superior precision in modulating voluntary activation. These differences likely reflect asymmetries in descending corticospinal drive, inhibitory regulation, and motor unit recruitment strategies, which may contribute to the greater susceptibility of elbow extensors to functional impairment following SCI.
The third study compares two rehabilitation approaches for persistent neck pain: a motion-driven BoMI with virtual reality (VR) feedback versus conventional therapist-guided training. Participants in both groups completed six weeks of sensorimotor training, with the VR group using motion tracking and real-time visual feedback to guide cervical movements during gamified tasks. While both groups showed improvements in pain and disability, the BoMI-guided VR group exhibited greater enhancements in cervical motion quality and coordination, as quantified through kinematic and finite helical axis analyses. These findings underscore the added value of BoMI-based VR rehabilitation in promoting functional recovery and eliciting neuromechanical adaptations not observed with conventional therapy.
Overall, these studies demonstrate that sensor-based BoMIs can be effectively applied to address key challenges in motor rehabilitation, from guiding selective muscle activation and assessing neuromuscular control to supporting recovery in clinical populations. Across distinct contexts, BoMIs enabled real-time, task-specific interaction driven by biosignals, allowing both functional engagement and deeper analysis of motor performance. By integrating biosignal-based control with physiologically informed task design, this thesis contributes to advancing the use of BoMIs as practical tools for rehabilitation and as experimental platforms for investigating motor control in both neurotypical and clinical populations.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|