| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 5.45 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
A neurofibromatose tipo 1 (NF1) é uma doença hereditária associada a perturbações do desenvolvimento neurológico sendo um dos impactos desta doença as alterações da neuroplasticidade, incluindo a do córtex visual. Surge então a necessidade de aferir a integridade da neuroplasticidade, através dos potenciais evocados visuais (VEP) sob a forma de potenciação da resposta seletiva a estímulos (SRP). A SRP está presente nos VEP quando o potencial obtido tem maior intensidade num estímulo familiar do que num estímulo novel. Assim, o principal objetivo deste trabalho foi a classificação de estímulos novel e familiar presentes no eletroencefalograma (EEG), após cada estímulo. Foram então estudados métodos que conseguissem extrair do sinal EEG características diferenciadoras de cada tipo de estímulo que permitissem a sua classificação. Para atingir este objetivo foi então desenvolvido um programa com recurso a técnicas de Machine Learning que dado um período temporal após cada estímulo fosse capaz de classificar o estímulo que resultou na resposta presente no período temporal, como novel ou familiar. Com o presente trabalho, foi possível concluir que é possível um algoritmo de machine learning classificar corretamente cada estímulo, uma vez que pelos resultados obtidos os valores da accuracy são bastante bons e o modelo apresenta robustez.
Neurofibromatosis type 1 (NF1) is an inherited disease associated with neurodevelopmental disorders, and one of the impacts of this disease is neuroplasticity changes, including that of the visual cortex. The need arises to assess the integrity of neuroplasticity, through visual evoked potentials (VEP) in the form of stimulus-selective response plasticity (SRP). SRP is present in VEP when the obtained potential has higher intensity in a familiar stimulus than in a novel stimulus. Thus, the main goal of this work was the classification of novel and familiar stimuli present in the electroencephalogram (EEG) after each stimulus. Methods were then studied that could extract from the EEG signal differentiating characteristics of each stimulus type that would allow their classification. To achieve this goal a program using techniques of Machine Learning was then developed, that given a time period after each stimulus was able to classify the stimulus that resulted in the response present in the time period, as novel or familiar. With this work, it was possible to conclude that it is possible for a machine learning algorithm to correctly classify each stimulus, since by the results obtained the values of accuracy are quite good and the model presents robustness.
Neurofibromatosis type 1 (NF1) is an inherited disease associated with neurodevelopmental disorders, and one of the impacts of this disease is neuroplasticity changes, including that of the visual cortex. The need arises to assess the integrity of neuroplasticity, through visual evoked potentials (VEP) in the form of stimulus-selective response plasticity (SRP). SRP is present in VEP when the obtained potential has higher intensity in a familiar stimulus than in a novel stimulus. Thus, the main goal of this work was the classification of novel and familiar stimuli present in the electroencephalogram (EEG) after each stimulus. Methods were then studied that could extract from the EEG signal differentiating characteristics of each stimulus type that would allow their classification. To achieve this goal a program using techniques of Machine Learning was then developed, that given a time period after each stimulus was able to classify the stimulus that resulted in the response present in the time period, as novel or familiar. With this work, it was possible to conclude that it is possible for a machine learning algorithm to correctly classify each stimulus, since by the results obtained the values of accuracy are quite good and the model presents robustness.
Description
Keywords
Classificação Eletroencefalograma Machine Learning Neuroplasticidade Potenciação da Resposta Seletiva a Estímulos Potenciais Evocados Visuais Classification Electroencephalogram Machine Learning Neuroplasticity Stimulus-Selective Response Potentiation Visual Evoked Potentials