Título: Feature Extraction from EEG signals for detection of Parkinsons Disease
Autores: Carolline Angela dos Santos Souza, Giovanni Guimaraes Viana, Bruno Fonseca Oliveira Coelho, Ana Beatriz Rodrigues Massaranduba, Rodrigo Pereira Ramos
Resumo: The Electroencephalogram (EEG) is a medical tool that captures, in a non-invasive way, electrical signals from the brain activities performed by neurons. EEG signals have been the target of study as a biomarker of Parkinsons disease (PD), where several methods of analysis are applied. The present work aims to evaluate features extracted from EEG signals, through methodologies such as HOS, Haralick descriptors, and Fractal Features, as new biomarkers for PD identification. Data from 50 individuals, available at the Open Neuro repository, who underwent an attentional cognitive task were analyzed. RF and SVM algorithms were employed for the classification of the extracted features. The best accuracy achieved was 79.49% in differentiating between Parkinsons subjects and control subjects using Haralick descriptors and RF classifier, suggesting that these features can identify activations in brain areas caused by dopaminergic medication
Palavras-chave: Parkisons Disease. Features. Haralick. Fractals
Páginas: 6
Código DOI: 10.21528/CBIC2023-027
Artigo em pdf: CBIC_2023_paper027.pdf
Arquivo BibTeX: CBIC_2023_027.bib