People suffering from upper limb impairments after a stroke or spinal cord injury are not only restricted in their independence but also in their inclusion in professional and social life. The increasing number of patients and the resulting rise in timely and monetary rehabilitation expenses lead to strong demands for new, effective therapies. Neuroprostheses based on functional electrical stimulation (FES) have been found to influence motor recovery positively. Electrical pulses are applied to peripheral nerves in the forearm and hand to generate functional hand motions. However, noninvasive hand neuroprostheses (HNPs) for rehabilitation face several challenges in clinical practice. The limited selectivity of transcutaneous FES yields difficulties in achieving fine hand movements by stimulating the muscle-rich forearm. Intersubject variability in neuroanatomy and tolerance of the FES make an individual adjustment of spatial and temporal stimulation parameters obligatory. Furthermore, strategies are required for a quick and easy adaptation of stimulation parameters in real-time, as the neuromuscular system is subject to time-variant changes. In this thesis, new concepts and methods are presented on the road to a novel, adaptive HNP based on automation, closed-loop control, and user-centered design. The HNP features a new, modular hand sensor system for accurate real-time motion tracking of FES-induced movements. In contrast to glove-based approaches, the proposed solution maintains the sense of touch. Algorithms for measuring segment orientations, wrist and finger joint angles, and fingertip positions from up to 17 micro inertial sensors were developed for application in patients with severe motor impairment of the hand. The methods avoid extensive calibration movements performed by the patients and work robustly in magnetically disturbed environments, i.e., indoors. The sensor system was evaluated with four healthy subjects in different validation settings before it was applied in clinical studies. Selective and individual stimulation of hand motion was assured by utilizing electrode arrays for the HNP together with user-centered identification strategies. An effective search for suitable virtual electrodes, formed by multiple, active array elements, is essential for clinical acceptance and practicability of HNPs. Semi-automatic and automatic methods for identifying stimulation positions and intensities were developed, realizing different levels of user integration. The semi-automatic approach allows caregivers to continuously modify virtual electrodes via a touchscreen while the stimulation intensities are automatically controlled to achieve desired wrist extension. Both identification methods were evaluated in five stroke survivors and yield suitable stimulation setups for hand opening and closing in patients who could tolerate the FES, with the semi-automatic approach being 25% faster than the automatic. A static parameter setup throughout a therapy session does not account for changes in the muscular response. For example, the rotation of the forearm during reach-and-grasp tasks leads to a change in FES response due to the relative transition between the skin and underlying neuromuscular tissues. An automatic real-time adaptation strategy of virtual electrodes and stimulation intensity in electrode arrays was investigated for a secure grasp during forearm movements. The novel method facilitates dynamic repositioning of electrodes and optional closed-loop control of the stimulation intensity. The hand sensor system was used to estimate grasping strength when using elastic objects. Experiments in four able-bodied volunteers revealed that the automatic electrode adaptation generates a strong, stable grasp force regardless of the rotational state of the forearm, in contrast to static electrodes. In summary, the presented concepts and methods in this thesis contribute to a higher degree of automation and adaptation of HNPs, which in the long run will enhance the use of FES-based technology in rehabilitation and, thereby, promote the motor recovery of patients.


    Zugriff

    Download

    Verfügbarkeit in meiner Bibliothek prüfen


    Exportieren, teilen und zitieren



    Titel :

    Adaptive hand neuroprosthesis using inertial sensors for real-time motion tracking


    Weitere Titelangaben:

    Eine adaptive Handneuroprothese unter Verwendung von Inertialsensoren zur Bewegungserfassung in Echtzeit


    Beteiligte:

    Erscheinungsdatum :

    2020



    Medientyp :

    Sonstige


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Klassifikation :

    DDC:    629 / 610



    Upper-body movements: precise tracking of human motion using inertial sensors

    Sara Santos / Duarte Folgado / Hugo Gamboa | BASE | 2020

    Freier Zugriff

    Upper-body movements: precise tracking of human motion using inertial sensors

    Sara Santos / Duarte Folgado / Hugo Gamboa | BASE | 2020

    Freier Zugriff

    Modular finger and hand motion capturing system based on inertial and magnetic sensors

    Valtin, Markus / Salchow-Hömmen, Christina / Seel, Thomas et al. | DataCite | 2017

    Freier Zugriff

    Using Inertial Sensors in Driver Posture Tracking Systems

    Butnariu, Silviu / Mogan, Gheorghe / Antonya, Csaba | Springer Verlag | 2018


    Multiple sensors based hand motion recognition using adaptive directed acyclic graph

    Xue, Yaxu / Ju, Zhaojie / Xiang, Kui et al. | BASE | 2017

    Freier Zugriff