The pyramid representation of the functional network using resting-state fMRI. Issue 3 (12th November 2022)

• Record Type: Journal Article
• Authors: Yang, Zhipeng; Li, Luying; Peng, Yaxi; Qin, Yuanyuan; Li, Muwei
• Is Part Of: Psychoradiology. Volume 2:Issue 3(2022)
• Page Start: 99
• Page End: 111
• Publication Date: 2022-11-12
• Subjects: multi-scale -- functional network -- graph theory -- resting-state fMRI -- support vector machine -- Alzheimer's disease; Brain -- Imaging -- Periodicals; Nervous system -- Radiography -- Periodicals; 616.804757
• DOI: 10.1093/psyrad/kkac011
• Languages: English
• ISSNs: 2634-4416

Abstract

Abstract: Background: Resting-state functional magnetic resonance imaging (RS-fMRI) has been proved to be a useful tool to study the brain mechanism in the quest to probe the distinct pattern of inter-region interactions in the brain. As an important application of RS-fMRI, the graph-based approach characterizes the brain as a complex network. However, the network is susceptible to its scale that determines the trade-off between sensitivity and anatomical variability. Objective: To balance sensitivity and anatomical variability, a pyramid representation of the functional network is proposed, which is composed of five individual networks reconstructed at multiple scales. Methods: The pyramid representation of the functional network was applied to two groups of participants, including patients with Alzheimer's disease (AD) and normal elderly (NC) individuals, as a demonstration. Features were extracted from the multi-scale networks and were evaluated with their inter-group differences between AD and NC, as well as the discriminative power in recognizing AD. Moreover, the proposed method was also validated by another dataset from people with autism. Results: The different features reflect the highest sensitivity to distinguish AD at different scales. In addition, the combined features have higher accuracy than any single scale-based feature. These findings highlight the potential use of multi-scale features as markers of the disrupted topological organization in AD networks. Conclusion: We believe that multi-scale metrics could provide a more comprehensive characterization of the functional network and thus provide a promising solution for representing the underlying functional mechanism in the human brain on a multi-scale basis.