Hao Liu,1,* Zheng-Xue Gu,2,* Xiao-Tong Li,3 Xin Huang4 

1School of Ophthalmology and Optometry, Jiangxi Medical College, Nanchang University, Nanchang, Jiangxi, People’s Republic of China; 2Department of Radiology, Nanjing Central Hospital, Nanjing, People’s Republic of China; 3Queen Mary School, Jiangxi Medical College, Nanchang University, Nanchang, Jiangxi, People’s Republic of China; 4The Affiliated Eye Hospital, Jiangxi Medical College, Nanchang University, Jiangxi, People’s Republic of China

*These authors contributed equally to this work

Correspondence: Xin Huang, The Affiliated Eye Hospital, Jiangxi Medical College, Nanchang University, Jiangxi Province Key Laboratory of Ophthalmology and Vision Sciences, Jiangxi Clinical Research Center for Ophthalmic Disease, Jiangxi Provincial Key Laboratory of Vitreoretinal Diseases for Health, 463 Bayi Avenue, Nanchang, Jiangxi Province, 330006, People’s Republic of China, Tel +8615879215294, Email 334966891@qq.com

Background: Diabetic retinopathy (DR) is a prevalent microvascular complication of diabetes. Prior neuroimaging research has indicated that patients with DR exhibit diverse levels of disrupted brain function alongside a variety of ocular symptoms. Nevertheless, past investigations have predominantly focused on static brain activity changes, leaving uncertainties regarding the modifications in dynamic large-scale brain networks among DR patients.
Purpose: The aim of this study was to investigate the alterations in dynamic large-scale functional network connectivity in DR patients and its medical significance.
Methods: Forty-six patients with DR (type 2 diabetes mellitus) and 46 healthy controls, matched for age, gender, and education level, were enrolled in this study. Initial application of Independent Component Analysis (ICA) methods was used to extract the resting state network (RSN) from resting state functional magnetic resonance imaging (fMRI) data. Subsequently, sliding time window and k-means cluster analysis were employed to derive five stable repetitions of the dynamic functional network connectivity (dFNC) states and compare the differences in dFNC between the two cohorts for each state. Finally, the study investigated between-group variances in three dynamic temporal metrics.
Results: Significant between-group differences in dFNC were observed in states 1 and 2. Patients with DR, compared to healthy controls, exhibited reduced functional connectivity within the visual network (VN) and between the dorsal attention network (DAN) and VN, coupled with higher functional connectivity between the default mode network (DMN) and VN, cerebellum network (CN) and VN, and DMN-executive control network (ECN). Regarding the three dynamic temporal metrics, the study findings indicated that DR patients experienced a notable decline in the fraction of time and mean dwell time in state 1, while showing an increase in these metrics for state 3.
Conclusion: Our study reveals extensive dynamic functional network connectivity alterations among patients with DR, potentially linked to visual impairment and cognitive deficits. These discoveries offer valuable insights into the neural mechanisms that drive changes in dynamic large-scale brain networks in individuals with DR.

Keywords: diabetic retinopathy, functional magnetic resonance imaging, independent component analysis, resting state network