Utilizing graph Fourier transform for automatic Alzheimer’s disease detection from EEG signals

Alzheimer’s disease (AD) is rapidly increasing globally, poses a significant challenge in the field of neurodegeneration. A critical aspect is the development of biomarker tools to detect and monitor the disease’s progression, especially in its early stages. Electroencephalogram (EEG) signal analysis holds promise for automated Alzheimer’s diagnosis. However, existing methods frequently disregard the intricate functional interconnections within the brain during diverse activities, as they analyze each EEG channel in isolation. In response to these limitations, this research introduces a novel approach grounded in graph signal processing (GSP). In this paper, a new feature of graph Fourier transform (GFT) based on the concept of GSP has been introduced for the detection of Alzheimer. We have compared our work with the popular feature of discrete wavelet transform (DWT) and the existing features based methods. Our method provides 98.9% accuracy using decision tree (DT) classifiers on the Florida-based dataset, which is better than the existing state-of-art techniques for Alzheimer’s detection.