Graph convolutional networks for diagnosis of Alzheimer’s disease: a review

Abstract

Medical imaging is a common diagnostic tool for identifying Alzheimer's disease in patients. Magnetic resonance imaging provides structural and functional connectivity information about the brain in the form of a graph. Graphical data reside in non-Euclidean space and give information about the geometrical structure of data. Diagnosing neurological diseases requires the identification of key brain regions and their connections. In Alzheimer's, this information is present in the graph connectivity of brain regions. Therefore, graph neural networks have been trained on graphical data to diagnose Alzheimer's. In machine learning research, graph convolutional networks have been very popular in the last decade. Therefore, we present a survey on the application of graph convolutional networks for Alzheimer's disease. In this work, we present the contributions of prominent research papers in this area. We have searched and compiled recent works involving different variants of graph neural networks applied on Alzheimer's data. The research contributions are summarized in various sections with regard to model architecture, datasets, and classification of healthy, mild cognitively impaired, and Alzheimer's subjects. The research works are compared in terms of performance metrics such as accuracy, sensitivity, and specificity. This review also discusses the research gaps and possible future directions for researchers to improve the current state of the art. Copyright © 2026 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.