Browsing by Author "Carreres, Alicia"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
- Classification of alzheimer’s electroencephalograms using artificial neural networks and logistic regressionPublication . Rodrigues, Pedro Miguel; Teixeira, João Paulo; Hornero, Roberto; Poza, Jesús; Carreres, AliciaThe Artificial Neural Networks have been used over the years to solve complex problems and their development has strongly grown in recent years. In particular, this work, focused on the development and a comparison between Artificial Neural Networks (ANN) and a traditional statistical technic known as Logistic Regression (LR) in Electroencephalogram (EEG) classification. The Wavelet Transform was seen as the main technique of signal processing, in order to analyze the EEG signals of this study. Some features were extracted by the EEG signals like relative power (RP) in conventional frequency bands and two spectral ratios. The best feature combination was selected by Principal Components Analysis method to increase the accuracy of the ANN and LR to discriminate their entries between Alzheimer Disease and Controls.
- Evaluation of EEG spectralfeatures in Alzheimer disease discriminationPublication . Rodrigues, Pedro Miguel; Bispo, Bruno; Freitas, Diamantino Silva; Teixeira, João Paulo; Carreres, AliciaAlzheimer’s disease (AD) is considered one of the most disabling diseases and it has a high prevalence in developed countries. It is as well the most common cause of dementia and it affects particularly the elderly. The current AD diagnosis accuracy is relatively low. It is therefore necessary to optimize the methods for AD detection. The electroencephalogram (EEG) is an inexpensive and noninvasive technique, that is able to record the electromagnetic fields produced by the brain activity. It has shown in the recent past a growing quality of the contribution to show brain disorders. The aim of this study was to evaluate the individual and combined power of several EEG features in AD discrimination. 95.00% of sensitivity, 100.00% of specificity, 97.06% of accuracy and 0.98 of AUC were the best classification results obtained in this work.