Melhoria da eficiência da estimação de frequências de potenciais evocados visuais de estado estacionário utilizando métodos paramétricos de alta resolução

• Autor(a) principal: Magalhães, Marcela Gonçalves
• Data de Publicação: 2019
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFG
• dARK ID: ark:/38995/0013000001sj0
• Texto Completo: http://repositorio.bc.ufg.br/tede/handle/tede/9465
• topic: EEG; Estimação de frequências; Potenciais evocados visuais; Métodos paramétricos; EEG; Frequency estimation; Visual evoked potentials; Parametric methods; ENGENHARIAS::ENGENHARIA ELETRICA

This work presents an analysis of electroencephalographic signal processing focusing on steady- state visual evoked potentials. It is sought to improve the estimation efficiency of the brain signal frequencies of the occipital region of the scalp that are produced in response to arbitrary visual stimuli. For this, we first use synthetic sinusoidal signals and then electroencephalographic signals obtained from an open-access public database. For the estimation, we used the prediction error filter applied in an adaptive way through the LMS algorithm and also developed versions of the MODE and MODEX methods for frequency estimation. These methods were compared with the classical approach of the problem, which uses the Fast Fourier Transform, in terms of root mean square error estimation for different values of signal-to-noise ratio, computational effort and execution time. The adaptive filtering and the worked estimators presented estimation errors lower than those of the classical approach for SNRs above 5 dB. Although the complexity and execution times of the proposed approaches have presented higher values than those of the FFT, they are partially offset by the reduction in the time required for data acquisition, so that the increase in accuracy and robustness is worth the increase in processing effort. Finally, it can be concluded that DOA estimation methods can be applied successfully in frequency estimation, particularly for EEG signals, allowing the development of more sophisticated applications in the future.