ViT-CNN: Leveraging a hybrid convolutional neural network and vision transformer for Alzheimer’s disease classification based on EEG signal

Alzheimer's disease is a prevalent and progressive neurodegenerative condition, primarily associated with cognitive impairments in aging populations. At present, there is no established standardized treatment for Alzheimer’s. As a result, the treatment strategy focuses on identifying key indicators for timely detection and preventing the progression of the condition. EEG is a valuable tool for identifying alterations in brain function during the early phases of Alzheimer's. Machine learning and deep learning methods can be applied to analyze EEG signals based on the detection or non-detection of Alzheimer's. These methods extract significant characteristics from the brainwave data. This article presents a new framework for classifying Alzheimer’s based on EEG signals. At first, EEG signals are pre-processed and converted to spectral images. Then a vision transformer and a hybrid method based on convolutional neural network and vision transformer are utilized for classification. The performance of the proposed models is assessed using different performance measures. The results show the model's effective classification of the time-frequency representations of EEG signals, exhibiting high accuracy and balanced performance across all metrics.

کلیدواژه ها:

Alzheimer’s disease ، Electroencephalography ، deep learning ، vision transformer ، Residual Neural Network