Deep learning methods for classification Alzheimer's disease

Alzheimer's disease (AD) is a progressive and irreversible brain disorder that affects memory, thinking, and behavior. It is the leading cause of dementia. Early diagnosis can slow the progression of Alzheimer's disease and improve the prognosis and increase the quality and quantity of patient care. One of the primary methods in early diagnosis is electroencephalography (EEG), which has been indicated as a promising method for detecting aberrant brain patterns associated to Alzheimer's disease in aspects of low cost, noninvasive, and portability. Furthermore, artificial intelligence tools have been essential in developing models that facilitate disease diagnosis and detection. Deep learning is a promising approach for such applications; however, it requires a reliable dataset. Due to the patient's rights, researchers may not be able to access a sufficient dataset to train the network. This work aims to propose a model to address this issue. Frist, Generative Adversarial Networks (GNN) model is presented to generate an artificial EEG dataset for Alzheimer's disease. It may be employed to understand brain processes better and make more accurate medical diagnoses for Alzheimer's disease using deep learning tools. Then four models have been focused on, Convolutional neural Network (CNN), Recurrent Neural Network (RNN), Multi-layer perceptron (MLP) and Transformer, to classify the Alzheimer's EEG signals. The results show that the GNN model can generate reliable artificial EEG signals for Alzheimer's disease in related channels. Moreover, the proposed models achieve accurate classification with a high accuracy 99.98 %, 99.76 %, 97.58 %, and 97.34 % respectively. Our study has demonstrated that the proposed methodologies serve as a promising complementary tool for identifying potential biomarkers that can aid in the clinical diagnosis of Alzheimer's disease. Keywords: Alzheimer's Disease, Electroencephalography (EEG), Generative Adversarial Network (GAN), Convolutional neural Network (CNN), Recurrent Neural Network (RNN), Multi-layer perceptron (MLP), and Transformer