Investigation of the correlation between brain functional connectivity and ESRD based on low‐order and high‐order feature analysis of rs‐fMRI

Peirui Bai,Yulong Wang,Feng Zhao,Qingyi Liu,Chengjian Wang,Jun Liu,Yaqian Qiao,Chi Ma,Yande Ren

DOI: https://doi.org/10.1002/mp.16410

IF: 4.506

2023-04-07

Medical Physics

Abstract:Background : The lack of analysis of brain networks in individuals with end‐stage renal disease (ESRD) is an obstacle to detecting and preventing neurological complications of ESRD. Purpose : This study aims to explore the correlation between brain activity and ESRD based on a quantitative analysis of the dynamic functional connectivity (dFC) of brain networks. It provides insights into differences in brain functional connectivity between healthy individuals and ESRD patients and aims to identify the brain activities and regions most relevant to ESRD. Methods : Differences in brain functional connectivity between healthy individuals and ESRD patients were analyzed and quantitatively evaluated in this study. Blood oxygen level‐dependent (BOLD) signals obtained through resting‐state functional magnetic resonance imaging (rs‐fMRI) were used as information carriers. First, a connectivity matrix of dFC was constructed for each subject using Pearson correlation. Then a high‐order connectivity matrix was built by applying the "correlation's correlation" method. Second, sparsification of the high‐order connectivity matrix was performed using the graphical least absolute shrinkage and selection operator (gLASSO) model. The discriminative features of the sparse connectivity matrix were extracted and sifted using central moments and t‐tests, respectively. Finally, feature classification was conducted using a support vector machine (SVM). Results : The experiment showed that functional connectivity was reduced to some degree in certain brain regions of ESRD patients. The sensorimotor, visual, and cerebellum subnetworks had the highest numbers of abnormal functional connectivities. It is inferred that these three subnetworks most likely have a direct relationship to ESRD. Conclusions : The low‐order and high‐order dFC features can identify the positions where brain damage occurs in ESRD patients. In contrast to healthy individuals, the damaged brain regions and the disruption of functional connectivity in ESRD patients were not limited to specific regions. This indicates that ESRD has a severe impact on brain function. Abnormal functional connectivity was mainly associated with the three functional brain regions responsible for visual processing, emotional, and motor control. The findings presented here have potential for use in the detection, prevention, and prognostic evaluation of ESRD. This article is protected by copyright. All rights reserved

radiology, nuclear medicine & medical imaging