2025-12-022025-12-022025-08-22FAVA, Maria Fernanda de Leão. Classificação de sinais EEG em Indivíduos com TEA utilizando redes neurais convolucionais e espectros HHSA. Orientadora: Suzana Cescon de Souza. 2025. 50f. Trabalho de Curso (Bacharelado em Engenharia Elétrica e Biomédica) – Faculdade de Engenharia Elétrica e Biomédica, Instituto de Tecnologia, Universidade Federal do Pará, Belém, 2025. Disponível em: https://bdm.ufpa.br/handle/prefix/8920 . Acesso em:.https://bdm.ufpa.br/handle/prefix/8920The work investigates the classification of EEG signals electroencephalogram) of individuals with autistic spectrum disorder (ASD) using convolutionary neural networks (CNN) and spectral analysis of Holo-Hilbert (HHSA). Research aims to overcome the limitations of current diagnostic methods- mostly behavioral and subjective- through a noninvasive and objective computational approach. The data, obtained from 12 participants (6 ASD and 6 Control), were converted to three-dimensional spectral maps representing the frequency and amplitude modula tion in brain channels (P3 and P4). These maps were used as an entry for a personalized CNN, trained with regularization techniques such as Dropout, Batch Normalization, Data Augmentation and Early Stopping. CNN performed promising performance even with limited database, with an average accuracy of 82,99% and AUC of 0,92 in cross validation. The results indicate the viability of the use of advanced spectrum-time representations combined with deep learning to assist in early and more accurate diagnosis of ASD.Acesso AbertoTranstorno do espectro autistaEletroencefalogramaAnálise holo-hilbertRedes neurais convolucionaisClassificação automáticaAutistic spectrum disorderElectroencephalogramHolo-hilbert analysisConvolutionary neural networksAutomatic classificationCNPQ::ENGENHARIAS::ENGENHARIA BIOMEDICATrabalho de Curso - Graduação - MonografiaAttribution-NonCommercial-NoDerivs 3.0 Brazil