Edge-centric effective connection network based on muti-modal MRI for the 
diagnosis of Alzheimer’s disease

Zhang, Shunqi; Zhao, Haiyan; Wang, Weiping; Wang, Zhen; Luo, Xiong; Hramov, Alexander; Kurths, Jürgen

doi:10.1016/j.neucom.2023.126512

アイテム詳細

登録内容を編集ファイル形式で保存

一時保存へ追加

タグ情報を表示リリース履歴を表示詳細要約

公開

学術論文

Edge-centric effective connection network based on muti-modal MRI for the diagnosis of Alzheimer’s disease

Authors

Zhang, Shunqi
External Organizations;

Zhao, Haiyan
External Organizations;

Wang, Weiping
External Organizations;

Wang, Zhen
External Organizations;

Luo, Xiong
External Organizations;

Hramov, Alexander
External Organizations;

/persons/resource/Juergen.Kurths

Kurths, Jürgen
Potsdam Institute for Climate Impact Research;

URL

There are no locators available

フルテキスト (公開)

There are no public fulltexts stored in PIKpublic

付随資料 (公開)

There is no public supplementary material available

引用

Zhang, S., Zhao, H., Wang, W., Wang, Z., Luo, X., Hramov, A., & Kurths, J. (2023). Edge-centric effective connection network based on muti-modal MRI for the diagnosis of Alzheimer’s disease. Neurocomputing, 552:. doi:10.1016/j.neucom.2023.126512.

引用: https://publications.pik-potsdam.de/pubman/item/item_28934

要旨

Alzheimer’s disease (AD) is an irreversible neurodegenerative disease. But if AD is detected early, it can greatly reduce the severity of the disease. Functional connection networks (FCNs) can be used for the early diagnosis of AD, but they are undirected graphs and lack the description of causal information. Moreover, most of FCNs take brain regions as nodes, and few studies have been carried out focusing on the connections of the brain network. Although effective connection networks (ECNs) are digraphs, they do not reflect the causal relationships between brain connections. Therefore, we innovatively propose an edge-centric ECN (EECN) to explore the causality of the co-fluctuating connection in brain networks. Firstly, the traditional conditional Granger causality (GC) method is improved for constructing ECNs based on the suppression relationship between structural connection network (SCN) and FCN. Then based on the improved GC method, edge time series and EECNs are constructed. Finally, we perform dichotomous tasks on four stages of AD to verify the accuracy of our proposed method. The results show that this method achieves good results in six classification tasks. Finally, we present some brain connections that may be essential for early AD classification tasks. This study may have a positive impact on the application of brain networks.