Detecção da doença de Alzheimer em sinais de EEG utilizando aprendizado profundo

Abstract

Alzheimer’s Disease is a progressive neurodegenerative disorder that affects cognitive functions such as memory, attention, and language, and whose early diagnosis still represents a relevant clinical challenge. Among the assessment methods, the electroencephalogram (EEG) stands out as a non-invasive technique capable of capturing changes in the brain’s electrical dynamics associated with dementia conditions. However, the manual analysis of these signals is complex and subject to variability among specialists, which motivates the use of advanced computational techniques for diagnostic support. This work investigates the possibility of using deep learning to assist in the detection of Alzheimer’s and to distinguish different clinical conditions through EEG signals. The research proposes a methodology based on the Discrete Wavelet Transform (DWT) to represent the time-frequency characteristics of the EEG segments, followed by the development of a convolutional neural network (CNN) model for multiclass classification among individuals with Alzheimer’s, Frontotemporal Dementia, and healthy controls.