Şen, Sena YağmurCura, O.K.Akan, Aydın2023-12-262023-12-2620239798350311402https://doi.org/10.1109/CoDIT58514.2023.10284052https://hdl.handle.net/20.500.14365/5032IEEE;LISIER;Sapienza Universita di Roma9th International Conference on Control, Decision and Information Technologies, CoDIT 2023 -- 3 July 2023 through 6 July 2023 -- 193871Dementia is a prevalent neurological disorder that results in cognitive function decline, significantly impacting the quality of life. In this study, a signal decomposition based method is proposed for the detection and follow-up Alzheimer's Dementia (AD) by using Electroencephalography (EEG) signals. The proposed approach uses the Intrinsic Time Scale Decomposition (ITD) to classify EEG segments of AD patients and control subjects. Signal decomposition process is conducted with 5 seconds EEG segment duration. Proper Rotation Components (PRCs) extracted from the EEG segments are used to train a 1-Dimensional Convolutional Neural Network (1D CNN). The proposed method is compared with classification of 5s duration EEG segments using the same CNN architecture. The experimental results demonstrate that utilizing ITD based approach yields better classification performance when compared to using the plain EEG signals. © 2023 IEEE.eninfo:eu-repo/semantics/closedAccessBiomedical signal processingConvolutional neural networksDeep learningElectrophysiologyNeurodegenerative diseasesAlzheimer dementiaCognitive functionsControl subjectDecomposition processDementia patientsFollow upIntrinsic time-scale decompositionsNeurological disordersQuality of lifeSignal decompositionElectroencephalographyConference Object10.1109/CoDIT58514.2023.102840522-s2.0-85177469825