Publicação
Goal-directed reaching arm movements: an EEG study
| Resumo: | The natural control of neuroprostheses is currently a challenge in both rehabilitation engineering and brain-computer interfaces (BCIs) research. One of the recurrent problems is to know exactly when to activate such devices. For the execution of the most common activities of daily living, these devices only need to be active when in the presence of a goal. For that reason, and since goals drive our motor behaviors, the distinction between the planning of goal-directed and aimless movements, using non-invasive recordings, can be useful for the implementation of a simple and effective activation method for these devices. In this thesis, the differences between goal-directed and aimless movements were investigated, using high-density electroencephalography (EEG) measurements in 10 healthy subjects, in a reach-and-touch task. Event-related potentials and oscillatory activity changes were studied. Furthermore, different feature extraction methods are used to train linear classifiers and test offline classification in a single-trial basis. The results show that there are statistically significant differences between the ERP waveforms of both Goal and No-Goal tasks, specifically in the N200-P300 complex. Furthermore, those differences are possible to classify using linear discriminant analysis, reaching average accuracies higher than 80%, and per-subject accuracies above 90% depending on the features and channels used for classification. In respect to the ERD/ERS phenomena, the upper alpha rhythm suppression is more prolonged and stronger with the goal-directedness of the task in the electrodes over the posterior parietal cortex. Furthermore, it was clear that distinct beta ERS waveforms were observed in contralateral central electrodes after movement execution. The synchronization in the middle beta band was stronger after the execution of aimless movements. These promising results could be used for new BCIs implementation. Combining the results in this study with movement decoding would be useful to achieve a more natural control strategy for BCIs, exclusively relying on the cognitive processes behind movement preparation and execution. |
|---|---|
| Autores principais: | Pereira, Joana Ministro dos Santos Caldas |
| Assunto: | Interfaces cérebro-computador/máquina EEG Potenciais evocados Sincronização/dessincronização Movimentos orientados por objectivo Planeamento de movimento N200 P300 Análise tempo-frequência Teses de mestrado - 2015 |
| Ano: | 2015 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório da Universidade de Lisboa |
| Resumo: | The natural control of neuroprostheses is currently a challenge in both rehabilitation engineering and brain-computer interfaces (BCIs) research. One of the recurrent problems is to know exactly when to activate such devices. For the execution of the most common activities of daily living, these devices only need to be active when in the presence of a goal. For that reason, and since goals drive our motor behaviors, the distinction between the planning of goal-directed and aimless movements, using non-invasive recordings, can be useful for the implementation of a simple and effective activation method for these devices. In this thesis, the differences between goal-directed and aimless movements were investigated, using high-density electroencephalography (EEG) measurements in 10 healthy subjects, in a reach-and-touch task. Event-related potentials and oscillatory activity changes were studied. Furthermore, different feature extraction methods are used to train linear classifiers and test offline classification in a single-trial basis. The results show that there are statistically significant differences between the ERP waveforms of both Goal and No-Goal tasks, specifically in the N200-P300 complex. Furthermore, those differences are possible to classify using linear discriminant analysis, reaching average accuracies higher than 80%, and per-subject accuracies above 90% depending on the features and channels used for classification. In respect to the ERD/ERS phenomena, the upper alpha rhythm suppression is more prolonged and stronger with the goal-directedness of the task in the electrodes over the posterior parietal cortex. Furthermore, it was clear that distinct beta ERS waveforms were observed in contralateral central electrodes after movement execution. The synchronization in the middle beta band was stronger after the execution of aimless movements. These promising results could be used for new BCIs implementation. Combining the results in this study with movement decoding would be useful to achieve a more natural control strategy for BCIs, exclusively relying on the cognitive processes behind movement preparation and execution. |
|---|