PPGINF PÓS-GRADUAÇÃO EM ENGENHARIA DA INFORMAÇÃO FUNDAÇÃO UNIVERSIDADE FEDERAL DO ABC Phone: Not available http://propg.ufabc.edu.br/ppginfo

Banca de QUALIFICAÇÃO: DOUGLAS CANONE GARCIA

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
DISCENTE : DOUGLAS CANONE GARCIA
DATA : 01/06/2022
HORA: 15:00
LOCAL: por participação remota
TÍTULO:

Hybrid Brain-Computer Interfaces Based on Non-invasive Electrophysiology


PÁGINAS: 79
GRANDE ÁREA: Engenharias
ÁREA: Engenharia Biomédica
SUBÁREA: Bioengenharia
ESPECIALIDADE: Processamento de Sinais Biológicos
RESUMO:

Expanding the integration between humans and assistive technology devices drives the demand for the development of techniques that allow greater independence for people affected by motor and sensory limitations, caused by accidents or advancing age. The concept of Brain Computer Interface (BCI) proposes a development of assistive technology devices using the analysis of biosignals from the brain and its implementation challenges increase as less as invasive procedures are adopted for obtain brain signals. The Hybrid Brain Computer Interface (hBCI) concept emerges as an alternative to make viable BCI systems, and this is possible through the fusion of biosignals from other parts of the body to compose a hybrid system that maximizes results. This work begins by presenting a brief literature review of scientific articles that fit within this concept of hBCI, but with an additional constraint of non-invasive electrophysiological signals obtained through surface electrodes. Following, the results and discussion of a study involving real movement classification for hBCI applications are presented, which aimed to identify and classify electroencephalographic (EEG) and electromyographic (EMG) data recorded during the execution of real movements, which were later stored and the resulting database, named by the authors as Multimodal Signal Dataset (MMSD), was publicly available on the internet. To validate the methodological approach adopted in this study, EEG data from the BCI Competition IV (dataset 2A) was used, obtaining a classification performance superior of that reached by the competition winners. The proposed methodology makes use of feature extraction and machine learning techniques well-known by the scientific community, although with some contributions and adjustments. In summary, the method uses Neighborhood Component Analysis to select regularized Common Spatial Pattern features extracted from EMG and EEG signals.  After validation, this same methodology is then applied to EMG and EEG recordings of the MMSD dataset, in a classification task consisting of differentiating three types of movement, involving different muscular groups.


MEMBROS DA BANCA:
Presidente - Interno ao Programa - 1545987 - FRANCISCO JOSE FRAGA DA SILVA
Membro Titular - Examinador(a) Interno ao Programa - 1946319 - DIOGO COUTINHO SORIANO
Membro Titular - Examinador(a) Interno ao Programa - 1544392 - ALINE DE OLIVEIRA NEVES PANAZIO
Membro Suplente - Examinador(a) Interno ao Programa - 1761107 - RICARDO SUYAMA
Notícia cadastrada em: 02/05/2022 09:30
SIGAA | UFABC - Núcleo de Tecnologia da Informação - ||||| | Copyright © 2006-2022 - UFRN - sigaa-2.ufabc.int.br.sigaa-2-prod