Publicação
Alzheimer’s disease recognition with artificial neural networks
| Resumo: | Alzheimer’s disease (AD) is the most common cause of dementia, and is well known for affect the memory loss and other intellectual disabilities. The Electroencephalogram (EEG) has been used as diagnosis tool for dementia over several decades. The main objective of this work was to develop an Artificial Neural Network (ANN) to classify EEG signals between AD patients and Control subjects. For this purpose two different methodologies and variations were used. The Short time Fourier transform (STFT) was applied to one of the methodologies and the Wavelet Transform (WT) was applied to the other methodology. The studied features of the EEG signals were the Relative Power in conventional EEG bands (delta, theta, alpha, beta and gamma) and their associated Spectral Ratios (r1, r2, r3 and r4). The best classification was performed by the ANN using the WT Biorthogonal 3.5 with AROC of 0.97, Sensitivity of 92.1%, Specificity of 90.8% and 91.5% of Accuracy. |
|---|---|
| Autores principais: | Rodrigues, Pedro Miguel |
| Outros Autores: | Teixeira, João Paulo |
| Assunto: | Artificial neural networks Wavelet transform Short time fourier transform Alzheimer’s disease Electroencephalogram |
| Ano: | 2013 |
| País: | Portugal |
| Tipo de documento: | capítulo de livro |
| Tipo de acesso: | acesso restrito |
| Instituição associada: | Instituto Politécnico de Bragança |
| Idioma: | inglês |
| Origem: | Biblioteca Digital do IPB |
| Resumo: | Alzheimer’s disease (AD) is the most common cause of dementia, and is well known for affect the memory loss and other intellectual disabilities. The Electroencephalogram (EEG) has been used as diagnosis tool for dementia over several decades. The main objective of this work was to develop an Artificial Neural Network (ANN) to classify EEG signals between AD patients and Control subjects. For this purpose two different methodologies and variations were used. The Short time Fourier transform (STFT) was applied to one of the methodologies and the Wavelet Transform (WT) was applied to the other methodology. The studied features of the EEG signals were the Relative Power in conventional EEG bands (delta, theta, alpha, beta and gamma) and their associated Spectral Ratios (r1, r2, r3 and r4). The best classification was performed by the ANN using the WT Biorthogonal 3.5 with AROC of 0.97, Sensitivity of 92.1%, Specificity of 90.8% and 91.5% of Accuracy. |
|---|