FenixEdu™

Dissertação

{en_GB=Using Brain-Computer Interface to understand and decode neural processes associated with imagined speech} {} EVALUATED

Detalhes: {pt=A fala imaginada desempenha um papel central na consciência humana, e a sua compreensão é extremamente importante para a construção duma Interface Homem-Máquina (IHM) semelhante à fala. Tais IHM ajudariam consideravelmente doentes incapazes de comunicação verbal, apesar de plenamente conscientes. Este estudo pretende construir uma IHM não-invasiva em circuito fechado utilizando uma tarefa baseada em vogais. Para tal, foi criada uma tarefa offline com vogais, sílabas e palavras para compreender melhor processos neuronais subjacentes à fala imaginada e avaliar exequibilidade de classificações binárias. Registámos o electroencefalograma (EEG) de seis indivíduos durante a tarefa offline (ouvir estímulos auditivos e subsequentemente imaginar dizê-los) e procedemos a uma análise comparativa entre diferentes abordagens para classificação. O melhor método consistiu na selecção semiautomática de features duma lista de 41 features espectrais de sub-bandas do EEG, seguida de análise de componentes principais e classificação segundo o centróide mais próximo. Através deste método, as exactidões médias de classificação de vogais versus repouso foram 68.9+/-12.4%d.p. (melhor indivíduo 85.5%) e 71.1+/-12.7% (90.6%) para as condições de fala imaginada e estimulação auditiva, respectivamente (p<0.05). Relativamente aos melhores pares de vogais, as correspondentes exactidões foram 58.7+/-6.0% (63.5%) e 60.4+/-3.9% (65.9%). A tarefa online, onde utilizadores controlavam uma bola num ecrã através de vogais imaginadas, resultou numa exactidão de 67.7+/-10.9% (83.9%) para o melhor par. Concluímos que IHM não-invasivas baseadas na fala imaginada estão ao nosso alcance mas que a sua aplicabilidade ainda requer esforços para melhorar aquisição e processamento de sinais e classificação., en=Imagined speech plays a central role in human consciousness, and its understanding is of great importance for the construction of a speech-like Brain-Computer Interface (BCI). Such BCIs would considerably benefit patients suffering from severe conditions where they are unable to verbally communicate, despite being fully conscious. The goal of this study was to build a non-invasive closed-loop BCI for a vowel task. To this end, an offline vowel / syllable / word task was designed to better understand neural processes underlying imagined speech and evaluate binary classification ability. We recorded six subjects’ brain activities using electroencephalography (EEG) during the offline task (listening to auditory stimuli and subsequently imagining saying them) and performed a comparative analysis between different classification approaches. The best method consisted of a semi-automatic feature selection from a list of 41 EEG sub-band spectral features, followed by principal component analysis and nearest centroid classification. Using this method, vowel vs rest average classification accuracies were 68.9+/-12.4%s.d. (best subject 85.5%) and 71.1+/-12.7% (90.6%) for imagined speech and auditory stimulation conditions, respectively (p<0.05). For the best pairs, vowel vs vowel corresponding average classification accuracies were 58.7+/-6.0% (63.5%) and 60.4+/-3.9% (65.9%). The online task, where users generated vowel speech imagery signals to control a ball on a screen, resulted in 67.7+/-10.9% (83.9%) classification accuracy for the best pair. We conclude non-invasive speech imagery BCIs are within reach, but their applicability still requires efforts in signal acquisition, processing and classification improvement, and in investigating the neural principles behind imagined speech.}
Keywords: {pt=Circuito fechado, Electroencefalografia, Fala Imaginada, Interface Homem-Máquina, Não-invasiva, en=Brain-Computer Interface, Closed-loop, Electroencephalography, Non-invasive, Speech Imagery}

Discussão: julho 31, 2015, 11:0