Development of a Neuro-Robot System for Lower-Limb Rehabilitation of Post-Stroke Patients Using Motorized Bike Pedal and Motor Imagery

Nome: Leticia Araújo Silva
Tipo: Dissertação de mestrado acadêmico
Data de publicação: 27/11/2020
Orientador:

Nome Papelordem decrescente
Teodiano Freire Bastos Filho Orientador

Banca:

Nome Papelordem decrescente
Eduardo Lázaro Martins Naves Examinador Externo
Denis Delisle Rodriguez Examinador Interno
Teodiano Freire Bastos Filho Orientador

Resumo: Stroke is a neurological syndrome that may affect severely upper and lower limbs movements, and the normal gait. The complete or partial restoration may be achieved through alternative rehabilitation therapies, such as Motor Imagery (MI)-based Brain-Computer Interfaces (BCIs). BCIs are becoming more popular for motor rehabilitation after stroke, due to their capability of translating the user’s brain waves into artificial output for controlling the external world. Although these systems have shown promising results on post-stroke patients with severe disability, their performance recognizing MI may be
reduced for people executing MI tasks with high difficult or producing weak brain activation. Some BCIs have been proposed to recognize motor imagery using different approaches for feature extraction, being Riemannian geometry, Common Spatial Pattern (CSP) and Filter Bank Common Spatial Pattern (FBCSP) very promising methods. This Master Dissertation aims to propose a new methodology and method for BCI training based on MI with pedal end-effector, which integrally aims to activate continuously the central and peripheral mechanisms related to lower-limbs, in order to provide practical
lower-limbs rehabilitation for post-stroke patients in clinical application. The proposed setup enables users to perform pedaling MI and receive passive pedaling into a calibration phase. Consequently, users can produce related Electroencephalogram (EEG) signals useful to obtain those more discriminant MI feature vectors through a feature vector analysis combining patterns from pedaling MI and real movements. Here, Riemannian geometry, CSP and FBCSP for feature extraction were used independently or combined in this approach. Also, four thresholds for feature vector analysis were evaluated, and three
different classifiers: Linear Discriminant Analysis (LDA) using diagonal matrix, LDA using full matrix and Regularized Fisher’s LDA (RFLDA). Accuracy (ACC), Kappa and relative power analysis are used to validated the Proposed BCI (pBCI). On calibration phase, the choice of threshold method demonstrated a high importance in ordtoer avoid underfitting. On online phase, the 75th percentile as threshold shows promising results when combined with Riemannian geometry, in the first feature extraction, and CSP, in the second feature extraction after threshold. For almost all participants was noted, during MI tasks, a power decreasing over the foot area (Cz location), corresponding to α (8 to 12 Hz) and β frequency bands, specifically for both low (13 to 22 Hz) and high (23 to 30 Hz) beta bands.

Acesso ao documento

Acesso à informação
Transparência Pública

© 2013 Universidade Federal do Espírito Santo. Todos os direitos reservados.
Av. Fernando Ferrari, 514 - Goiabeiras, Vitória - ES | CEP 29075-910