| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 6.86 MB | Adobe PDF |
Advisor(s)
Abstract(s)
Este trabalho tem como objetivo detectar a doença de Alzheimer em diferentes estágios através de sinais EEG. Potências relativas às frequências convencionais são obtidas dos valores máximo, mínimo e médio de estimativas do espectro de potências da Wavelet Packet. Para cada par de grupos de estudo e eletrodo, uma seleção dos parâmetros é realizada. Os parâmetros selecionados são utilizados como entrada para classificadores com validação cruzada leave-one-out. Acurácias de classificação de100% são obtidas, em pelo menos 1 eletrodo, para os 6 pares de grupos analisados, indicando as regiões no escalpe com maiores diferenças à medida que a doença progride. E acurácias médias, incluindo todos os eletrodos, entre 81,3 e 91,4% são alcançadas.
This work aims to detect the Alzheimer’s disease at different stages using EEG signals. Powers related to conventional frequencies are obtained from the maximum, minimum and average values of Wavelet Packet power spectrum estimates. For each pair of study groups and electrodes, a parameter selection is performed. The selected parameters are used as in put for classifiers with leave-one-out cross-validation. Classification accuracies of 100% are obtained, in at least 1 electrode, for the6 pairs of groups analyzed, indicating the regions in the scalp with the greatest differences as the disease progresses. Average classification accuracies, including all electrodes, between 81.3 and 91.4% are achieved.
This work aims to detect the Alzheimer’s disease at different stages using EEG signals. Powers related to conventional frequencies are obtained from the maximum, minimum and average values of Wavelet Packet power spectrum estimates. For each pair of study groups and electrodes, a parameter selection is performed. The selected parameters are used as in put for classifiers with leave-one-out cross-validation. Classification accuracies of 100% are obtained, in at least 1 electrode, for the6 pairs of groups analyzed, indicating the regions in the scalp with the greatest differences as the disease progresses. Average classification accuracies, including all electrodes, between 81.3 and 91.4% are achieved.
Description
Keywords
Doença de Alzheimer EEG Espectro de potência Sub-banda Wavelet packet Alzheimer disease Power spectrum Sub-band