Aberrant dynamic functional and effective connectivity changes of the primary visual cortex in patients with retinal detachment via machine learning

doi:10.21203/rs.3.rs-3808493/v1

Download PDF

Article

Aberrant dynamic functional and effective connectivity changes of the primary visual cortex in patients with retinal detachment via machine learning

https://doi.org/10.21203/rs.3.rs-3808493/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background: Retinal detachment (RD) is a prevalent and severe eye disease that often leads to vision loss. Previous research has indicated abnormal brain activity in individuals with RD. However, these studies solely focused on localized alterations in brain activity among individuals with RD, and it remains unclear if there are any changes in dynamic functional connectivity (dFC) and dynamic effective connectivity (dEC) in the primary visual cortex (V1) among individuals with RD.

Aim: This study utilizes seed-based functional connectivity (FC) analysis and Granger causality analysis (GCA) to examine alterations in dynamic functional and effective connectivity in the V1 among patients with RD.

Methods: The study involved 29 patients with RD and 30 healthy controls (HCs) who underwent resting-state functional magnetic resonance imaging (rs-fMRI) scans.Based on the seed regions in the V1, dynamic FC and GCA were conducted between the RD patients and HCs. To examine particular dFC and dEC states as well as associated temporal characteristics, the k-means clustering method was applied.The altered dFC and dEC values were selected as classification features and Support Vector Machine (SVM) classifiers were utilized to differentiate between patients with RD and HCs.

Results: Compared to HCs, patients with RD displayed a significantly increased dFC between the right V1 and the temporal lobe, thalamus, frontal lobe, occipital lobe, angular gyrus, and cerebellum.Additionally, patients with RD exhibited significantly increased dFC between the left V1 and the parietal lobe.On the other hand, patients with RD showed a significantly decreased dFC between the left V1 and the cerebellum, amygdala, temporal lobe, and frontal lobe.Using the dynamic GCA algorithm, patients with RD showed a significant increase in dEC outflow from the right V1 to the frontal lobe, the caudate, the parietal lobule, and the angular gyrus.Patients with RD also exhibited a significant increase in dEC inflow to the right V1 from the temporal lobe, thalamus, the occipital lobe, and the parietal lobe.Additionally, patients with RD had significantly increased dEC outflow from the left V1 to the frontal lobe and the parietal lobe.Furthermore, patients with RD displayed a significant increase in dEC inflow to the left V1 from the occipital lobe.In contrast, patients with RD showed a significant decrease in dEC outflow from the left V1 to the occipital lobe. Lastly, patients with RD had significantly decreased dEC inflow to the left V1 from the occipital lobe and the postcentral gyrus[two-tailed, voxel-level p < 0.05, Gaussian random field (GRF) correction, cluster-level p < 0.05].After performing k-means clustering, it was observed that patients with RD predominantly displayed three dFC states and three or four dEC states.Depending on the region of interest (ROI), there are differences in the number of transitions(NT), frequency(F), and mean dwell time(MDT).The SVM model demonstrated accuracies of 0.712, 0.695, 0.525, 0.542, 0.593, and 0.458, along with corresponding areas under the curve (AUC) of 0.729, 0.786, 0.492, 0.561, 0.572, and 0, respectively, in distinguishing between individuals with RD and HCs based on the dFC/dEC values for the different ROI.

Conclusion: Individuals with RD exhibited significant disruption in dFC/dEC between the V1 and multiple brain regions. The variability in dFC proved to distinguish individuals with RD from HCs with a high level of accuracy. These findings can contribute to the identification of potential neurological mechanisms underlying visual impairments in individuals with RD.

Health sciences/Medical research

Health sciences/Neurology

Health sciences/Diseases/Eye diseases

retinal detachment

resting-state functional magnetic resonance imaging

dynamic functional connectivity

dynamic effective connectivity

k-means clustering method

Support Vector Machine

Retinal detachment (RD) is defined as the separation of the neuroepithelial layer from the pigment epithelial layer in the retina. Failure to promptly address this condition can lead to necrosis of retinal photoreceptor cells and subsequent vision loss, as the detached retina does not receive a blood supply from the choroid. Common clinical symptoms associated with RD include the sudden onset of floaters, flashes of light, the presence of a dark shadow that obscures vision, and vision loss. Several factors contribute to RD, with diabetic retinopathy¹, high myopia², ocular trauma³, post-cataract surgery⁴, and the use of oral quinolone medications⁵ being the most prevalent causes. Various diagnostic methods are available to detect RD, including ocular B-ultrasound, optical coherence tomography (OCT), optomap, and mydriatic slit lamp examination of the fundus. However, these methods can only identify local eye lesions and are unable to detect abnormalities in the visual function area of the brain. Previous studies utilizing resting-state functional magnetic resonance imaging (rs-fMRI) have reported abnormal brain activity in regions associated with vision in individuals with RD^6–9. Nonetheless, it is still unknown whether there are specific abnormalities in the dynamic functional connectivity (dFC) and dynamic effective connectivity (dEC) related to the primary visual cortex (V1) in individuals with RD.

In recent years, rs-fMRI has emerged as a valuable noninvasive tool for studying neurological changes in the brain. It has the potential to uncover the underlying neural mechanisms associated with various diseases. One widely used metric in rs-fMRI analysis is functional connectivity (FC), which measures the correlation of blood-oxygen-level-dependent (BOLD) signaling between different brain regions¹⁰. By examining patterns of FC, researchers can gain insights into how different brain regions communicate and work together. However, FC alone cannot provide information about the causal relationship between brain regions; it merely indicates that two regions are correlated without revealing which region is causing the correlation. This is where Granger causality analysis (GCA) comes into play. GCA is a statistical method that utilizes multiple linear regression to explore the predictive power of past values from one time series on the current values of another time series¹¹. In the context of rs-fMRI, GCA can be applied to investigate the causal effects between brain regions, aiding in the identification of information flow directionality by analyzing the temporal dynamics of BOLD signals. Previous studies have integrated FC and GCA methods^12–14. However, recent research has shown that brain activity exhibits dynamic temporal changes¹⁵. Therefore, we incorporated sliding window analysis into the aforementioned methods to capture temporal dynamics and examine how brain activity and connectivity patterns change over time in individuals with RD. Additionally, the k-means clustering method, widely used for grouping data based on similarities, was applied to the dFC/dEC values of subjects across different sliding time windows. This facilitated the identification of distinct states or patterns in brain activity during the entire scanning duration. Finally, Support Vector Machine (SVM) is a well-recognized classification method in the field of machine learning¹⁶. SVM can efficiently process data in high-dimensional space and avoid overfitting, even with limited sample sizes¹⁷. As a result, SVM has found extensive application in the field of neuroimaging. Moreover, previous researchers have combined k-means clustering and SVM approaches to analyze dynamic interactions among the salience network (SN), central executive network (CEN), and default mode network (DMN) in patients with Alzheimer's disease¹⁸. In a recent study, abnormal FC and EC alterations were observed in patients with unilateral acute tinnitus and hearing loss. The study revealed abnormalities in FC/EC between the frontostriatal network and various nonauditory regions, as well as their relationship with tinnitus perception¹⁴. However, no studies to date have identified abnormalities in dFC and dEC in the V1 of patients with RD by integrating k-means clustering and SVM methods.

The objective of the study was to examine the changes in dFC and dEC between the V1 and other brain regions in patients with RD. Specifically, we performed a comprehensive analysis of specific connections involving V1 in these patients. Initially, V1 served as the seed region, and a sliding window approach was used to evaluate alterations in dFC and dEC between V1 and all other brain regions. Subsequently, we applied the k-means clustering method to classify the dFC/dEC values obtained for each subject, based on various seed regions, into distinct states. Finally, we employed SVM methods to determine whether abnormal dFC/dEC values could effectively distinguish RD patients from normal controls.This study proposed two hypotheses: (1) alterations in dFC/dEC values within V1 may shed light on potential neural mechanisms underlying visual function impairment in RD patients; and (2) as sensitive biomarkers, dFC/dEC levels can distinguish RD patients from healthy individuals.

2.1. Participants

The First Affiliated Hospital of Nanchang University's Ethics Committee gave its clearance for this study. Written informed consent was acquired from all participants. The study included 29 patients diagnosed with RD and 30 healthy controls (HCs). The inclusion criteria for RD patients were: (1) idiopathic RD involving one or both retinal tears, (2) RD affecting one or both quadrants, and (3) absence of any eye diseases in both eyes. Patients with recurrent RD, RD caused by high myopia, RD associated with eye trauma, RD caused by diabetes, history of fundus surgery, cardiovascular diseases, mental disorders, and cerebral infarction were excluded. The healthy controls in Nanchang City were randomly selected to match the age, sex, and educational background of the RD patients. The inclusion criteria for the healthy controls were: (1) absence of eye diseases and major diseases, (2) visual acuity better than 1.0, and (3) finishing a number of tests, such as an magnetic resonance imaging(MRI), an OCT, and an ocular B-scan ultrasound.

2.2. fMRI Data Acquisition

At the Department of Radiology in the First Affiliated Hospital, Nanchang University, China, rs-fMRI data were gathered using a 3T MR scanner (Siemens, Erlangen, Germany) that had an 8-channel phased-array head coil installed. The data acquisition involved capturing 240 resting-state volumes over an 8-minute period. The parameters used were as follows: field of view of 240 mm x 240 mm, repetition time (TR) of 2,000 ms, echo time of 40 ms, flip angle of 90°, matrix size of 64 x 64, slice thickness of 4 mm, and a 1 mm gap. Each brain volume consisted of thirty axial slices. The following three-dimensional MRI settings were used to obtain each participant's high-resolution T1-weighted images: TR of 1,900 ms, echo time of 2.26 ms, flip angle of 9°, field of view of 240 mm x 240 mm, matrix size of 256 x 256, 176 sagittal slices, and a slice thickness of 1 mm.

2.3. fMRI Data Preprocessing

The RESTplus 1.25 rs-fMRI data analysis tools was used to preprocess the data¹⁹. The preprocessing procedures listed below were used: (1) The data was categorized into structural and functional images. (2) The data format was converted from DICOM to NIFTI. (3) The first ten time points were eliminated to ensure data stability as they captured the unstable signals obtained when the subject entered the MRI scan. (4) Time layer correction was applied to align the time points of the compartmentalized scanning by the MRI machine. (5) Head movement correction was implemented, and subjects with head movements exceeding 3 mm and 3°were excluded. (6) Normalization was performed to standardize the individual space to the Montreal Neurological Institute (MNI) standardized space. (7) Image smoothing was conducted using a 6 mm (length, width, and height) smoothing kernel to reduce noise differences between neighboring brain regions, resulting in a smoother and more continuous signal and increasing the signal-to-noise ratio for improved sensitivity and reliability of subsequent analyses. (8) Detrending was performed to eliminate the upward drift in the signal value caused by machinery or external factors. (9) Regression of interfering covariates was carried out, including the utilization of the Friston-24 parameter model to account for head motion signal^20,21, global mean signal, white matter signal, and cerebrospinal fluid signal. (10) Temporal band-pass filtering was applied with a frequency range of 0.01 to 0.08 Hz to reduce noise, remove artifacts, and enhance the signal of interest.

2.4. Seed-based dFC analysis

The region of interest (ROI) in this study is the V1 domain of the brain, as shown in Fig. 1. The MNI coordinates for the V1 region, according to our previous article, are as follows: right V1 (8, -76, 10) and left V1 (-8, -76, 10)²². Using the Dynamic Brain Connectome (DynamicBC) toolbox v2.2²³,we created a spherical ROI with a radius of 6 mm using the V1 coordinates as the ROI. To analyze the data, we employed the sliding time window method, creating 201 time windows of 30 TR (60s) each, with a step length of 1 TR (2s). Within each window, We determined the correlation coefficients between the bilateral V1 and the average BOLD signal of other voxels in each subject's complete brain.To capture the change of these coefficients over time, we calculated the standard deviation (STD) of the Z-values for each voxel's correlation coefficients. This STD represents the coefficient of variation of dFC.

2.5. Blind Deconvolution Procedure

The BOLD signal serves as an indicator for measuring changes in neuronal activity within specific brain regions. It indirectly reflects the level of local neural activity by capturing variations in magnetic field properties around oxygenated and deoxygenated hemoglobin. Hence, the BOLD signal offers an indirect measure of neural activity²⁴. Recent studies have indicated that the modulation of the BOLD signal is influenced by cerebrovascular responses and neurovascular coupling²⁵. In order to address potential confounding effects presented by the hemodynamic response function (HRF), we employed techniques for Resting-State Hemodynamic Response Function Retrieval and Deconvolution. By using these methods, we were able to deconvolve the observed BOLD signal, which resulted in a more precise calculation of effective connectivity²⁶.

2.6. Seed-based dEC analysis

We applied the same method as the previous dFC analysis to define the ROI, and utilized the DynamicBC toolkit to perform dynamic seed-based GCA. The seed time series was denoted as X, while Y represented the temporal activity of voxels throughout the brain. The "X to Y" relationship described how activities in region X influenced activities in region Y, whereas the "Y to X" relationship represented the reciprocal influence. For the dynamic analysis, a sliding window approach with a window width of 30TR (60s) and a step length of 1TR (2s) was employed. Each subject's time series was divided into 402 windows, and the directed connection coefficient was calculated between the bilateral V1 region and each window, as well as the average BOLD signal of other voxels throughout the brain. Consequently, a series of sliding window correlation coefficients was obtained for each subject. To capture the temporal variations in these coefficients, we calculated the STD of the Z-values for the directed connection coefficients of each element. This STD represents the coefficient of variation of the directed connection coefficient.

2.7. Statistical analysis

The statistical analysis of STD FC and STD EC graphs was conducted in patients with RD and HCs using a two-sample t-test.To account for multiple comparison correction, Gaussian random field (GRF) methods were used with covariates for age and sex included in regression analysis(two-tailed, voxel level p < 0.05, GRF correction, cluster-level p < 0.05).

2.8. Clustering analysis

We utilized the k-means clustering algorithm to cluster the dFC/dEC values of the bilateral V1 regions in all subjects through a 30TR sliding window. The similarity between different time windows was assessed using the L1 distance function (Manhattan distance)²⁷ to ascertain the occurrence state of dFC/dEC. The object was to identify the repeated FC and EC patterns in each window through cluster analysis and determine the optimal clustering coefficient value, K. Lastly, we computed the characteristics of dFC/dEC, including the number of transitions(NT), frequency(F), and mean dwell time(MDT).

2.9. Support vector machine analysis

We employed the LIBSVM software package²⁸, which is based on Matlab 2017b, to perform SVM classification on individuals with RD and HCs. The main aim of this study was to explore the potential of utilizing dFC/dEC values as diagnostic indicators for RD. The SVM classifier utilized the radial basis kernel function (RBF) as the kernel function and evaluated the performance of different feature subsets through the leave-one-out cross-validation (LOOCV) technique. The analysis flowchart for this study is depicted in Fig. 2.

2.10.Validation analysis

To corroborate our findings regarding the variability of dFC/dEC obtained with sliding windows of 30 TR, we conducted confirmatory analyses utilizing sliding windows of 60 TR and 80 TR lengths.

3.1. Demographic characteristics

A cohort of 29 RD patients—14 females and 15 males—with a mean age of 53.862 ± 19.093 years were included in this study. In addition, 30 HCs with a mean age of 51.667 ± 13.922 years were present, 15 of whom were male and 15 of whom were female. The RD patients and HCs did not significantly differ in terms of age or gender. Table 1 provides specific demographic information.

Table 1

Features of the demographics of and RD patients and HCs
Characteristic	RD patients	HCs	P-value
Men/women	15/14	15/15	0.895^x
Age (years,mean ± STD)	53.862 ± 19.093	51.667 ± 13.922	0.615^t
Duration of detachment (days)	15 (7, 30)^a	N/A	N/A
IOP (mmHg)	15 (12, 16)^a	N/A	N/A
Axial length of eye (mm,mean ± STD)	24.390 ± 1.899	N/A	N/A
Corneal endothelial cell count (/mm²)	2422(1966,2545)^a	N/A	N/A
HAMA score	3 (1, 5)^a	N/A	N/A
Abbreviations: x, Data were obtained using Pearson’s chi-square tests;t ,Data were obtained using two-sample t-tests;a ,median(interquartile range);STD,standard deviation;RD, retinal detachment; HCs, healthy controls; IOP, intraocular pressure; HAMA, Hamilton Anxiety Scale; N/A, not applicable

3.2.Typical optomap, B-scan ultrasound, and OCT examinations of RD

Figure 3 presents representative images of optomap, B-scan ultrasound, and OCT examinations conducted on patients with RD. The images illustrate key indicators of RD, including: A) the presence of a grayish retinal eminence in the detached region, involving the macular area, with an identifiable hiatal hole situated above it; B) a distinct banded echo that corresponds to the spherical wall echo; and C) detachment of the retinal nerve epithelium.

3.3.Differences in dFC/dEC values between RD and HCs

Figure 4 illustrates the brain regions exhibiting significant differences in the dFC/dEC values of bilateral V1 between patients with RD and HCs using a sliding window of size 30TR. Compared to HCs, patients with RD displayed significantly increased dFC between the right V1 and the cerebellum vermis 9, the left superior temporal gyrus (L-STG), the right thalamus (R-THA), the left superior frontal gyrus medial (L-SFGmed), the right middle occipital gyrus (R-MOG), the left middle occipital gyrus (L-MOG), the left angular gyrus (L-ANG), the right middle frontal gyrus (R-MFG), and the right supramarginal gyrus (R-SMG). Additionally, patients with RD exhibited significantly increased dFC between the left V1 and both the left superior parietal gyrus (L-SPG) and the right superior parietal gyrus (R-SPG). On the other hand, patients with RD showed significantly decreased dFC between the left V1 and the right cerebellum crus Ⅰ, the right amygdala (R-AMYG), L-STG, and the right rectus (R-REC).Using the dynamic GCA algorithm, patients with RD showed a significant increase in dEC outflow from the right V1 to the right superior frontal gyrus orbital part (R-SFGorb), the left caudate (L-CAU), the right middle frontal gyrus orbital part (R-MFGorb), the R-MFG, the left middle frontal gyrus (L-MFG), the right inferior parietal lobule (R-IPL), and the right angular gyrus (R-ANG). Patients with RD also exhibited a significant increase in dEC inflow to the right V1 from L-STG, the left middle temporal gyrus (L-MTG), R-THA, the right calcarine (R-CAL), the left cuneus (L-CUN), R-MOG, and R-IPL. Additionally, patients with RD had significantly increased dEC outflow from the left V1 to the left superior frontal gyrus (L-SFG) and R-IPL.Furthermore, patients with RD displayed a significant increase in dEC inflow to the left V1 from the left precuneus (L-PCUN). In contrast, patients with RD showed a significant decrease in dEC outflow from the left V1 to the right precuneus (R-PCUN). Lastly, patients with RD had significantly decreased dEC inflow to the left V1 from R-PCUN and the left postcentral gyrus (L-PoCG) (Figure 4; Table 2;two-tailed, voxel level p < 0.05; GRF correction, cluster-level p < 0.05).

Table 2

Significant differences in dFC/dEC values between RD patients and HCs at a sliding window size of 30TR.
	Seed region	Direction	Brain region	BA	Peak t-score	MNI coordinates (x, y, z)	Cluster size (voxels)
dFC	R-V1	-	Cerebellum-vermis- 9	-	3.794	0,-57,-33	138
			L-STG	-	4.339	-48 ,0,-9	178
			R-THA	-	4.169	12,-9,9	91
			L-SFGmed	10	3.945	-3,54,6	99
			R-MOG	-	3.849	-9,-105,6	172
			L-MOG	19	4.171	-27,-96,18	104
			L-ANG	-	5.491	-45,-69,33	375
			R-MFG	-	3.844	36,6,45	190
			R-SMG	2	3.937	51,-30,42	343
	L-V1	-	R-Cerebellum-Crus-Ⅰ	-	-4.052	24,-90,-24	184
			R-AMYG	-	-4.047	24,-3,-15	138
			L-STG	-	-5.117	-48,-12,-3	106
			R-REC	-	-4.456	9,30,-18	94
			L-SPG	-	4.935	-33,-72,54	527
			R-SPG	-	3.909	33,-66,60	108
dEC	R-V1	X to Y	R-SFGorb	10	3.452	27,54,-3	83
			L-CAU	-	3.742	-6,6,3	151
			R-MFGorb	10	3.616	36,60,-3	62
			R-MFG	-	4.260	42,30,30	60
			L-MFG	-	3.325	-21,45,33	56
			R-IPL	-	4.110	36,-45,48	65
			R-ANG	-	3.634	39,-66,48	76
		Y to X	L-STG	-	4.182	-51,-27,3	59
			L-MTG	21	3.943	-66,-39,0	43
			R-THA	-	3.660	18,-15,18	43
			R-CAL	31	3.047	6,-69,15	30
			L-CUN	-	3.217	-3,-69,24	65
			R-MOG	-	3.593	33,-81,21	46
			R-IPL	-	3.197	39,-42,48	35
	L-V1	X to Y	L-SFG	-	3.677	-18,54,3	39
			R-PCUN	-	-3.795	12,-66,21	48
			R-IPL	40	3.764	42,-51,42	60
		Y to X	R-PCUN	-	-3.670	21,-57,24	30
			L-PoCG	4	-4.304	-63,-6,24	40
			L-PCUN	-	3.710	-12,-60,54	36
Abbreviations: dFC,dynamic functional connectivity;dEC, dynamic effective connectivity; HCs, healthy controls; RD, retinal detachment; BA, Brodmann area; MNI, Montreal Neurological Institute;TR,repetition time;R-V1,right primary visual area;L-V1,left primary visual area;X to Y,from seed region to whole brain;Y to X,from whole brain to seed region;L-STG,left superior temporal gyrus;R-THA, right thalamus;L-SFGmed, left superior frontal gyrus medial ; R-MOG, right middle occipital gyrus; L-MOG,left middle occipital gyrus; L-ANG, left angular gyrus ;R-MFG,right middle frontal gyrus;R-SMG,right supramarginal gyrus;R-AMYG,right amygdala;R-REC,right rectus;L-SPG,left superior parietal gyrus;R-SPG,right superior parietal gyrus;R-SFGorb,right superior frontal gyrus orbital part;L-CAU,left caudate;R-MFGorb,right middle frontal gyrus orbital part;L-MFG,left middle frontal gyrus;R-IPL,right inferior parietal lobule;R-ANG, right angular gyrus;L-MTG,left middle temporal gyrus; R-CAL,right calcarine;L-CUN,left cuneus;L-SFG,left superior frontal gyrus;R-PCUN,right precuneus;L-PCUN,left precuneus;L-PoCG,left postcentral gyrus (two-tailed, voxel-level p < 0.05, GRF correction, cluster-level p < 0.05).

3.4. Clustered dFC/dEC states

Figure 5 illustrates the utilization of the k-means clustering algorithm to cluster the dFC and dEC values of the bilateral V1 in all subjects within a 30TR sliding window.It was observed that patients diagnosed with RD predominantly exhibited three dFC states and three to four dEC states. Depending on the ROI, there are differences in the NT, F, and MDT.Compared to the HCs,significantly higher frequencies were observed in the RD group at state 4 in the R-V1 to whole-brain dEC clustering results (p = 0.005, t = 2.892).Further details are presented in Table 3.

Table 3

The temporal properties of dFC/dEC patterns RD patients and HCs at a sliding window size of 30TR.
	Seed region	Dynamic clustering index		P-value	t-value
dFC	R-V1	Number of transitions		0.661	-0.440
		Frequency	State1	0.908	0.117
			State2	0.537	0.622
			State3	0.486	-0.701
		Mean dwell time	State1	0.693	0.396
			State2	0.251	1.159
			State3	0.632	-0.482
	L-V1	Number of transitions		0.907	0.118
		Frequency	State1	0.935	0.082
			State2	0.975	0.031
			State3	0.913	-0.109
		Mean dwell time	State1	0.305	1.035
			State2	0.960	-0.051
			State3	0.329	-0.985
dEC	R-V1(X to Y)	Number of transitions		0.292	-1.064
		Frequency	State1	0.078	-1.797
			State2	0.330	-0.983
			State3	0.037	-2.133
			State4	0.005^#	2.892^#
		Mean dwell time	State1	0.820	-0.228
			State2	0.330	-0.983
			State3	0.075	-1.814
			State4	0.018	2.438
	R-V1(Y to X)	Number of transitions		0.747	-0.324
		Frequency	State1	0.313	1.017
			State2	0.330	-0.983
			State3	0.046	2.036
			State4	0.052	-1.982
		Mean dwell time	State1	0.313	1.017
			State2	0.330	-0.983
			State3	0.835	0.210
			State4	0.036	-2.149
	L-V1(X to Y)	Number of transitions		0.625	0.491
		Frequency	State1	0.376	-0.893
			State2	0.148	1.466
			State3	0.277	-1.097
			State4	0.390	-0.867
		Mean dwell time	State1	0.376	-0.893
			State2	0.673	0.424
			State3	0.146	-1.474
			State4	0.475	-0.719
	L-V1(Y to X)	Number of transitions		0.204	1.286
		Frequency	State1	0.197	1.301
			State2	0.248	-1.168
			State3	0.330	-0.983
		Mean dwell time	State1	0.237	1.195
			State2	0.084	-1.761
			State3	0.330	-0.983
Abbreviations: The number of transitions is evaluated using a two-sample t-test(p < 0.05), whereas the frequency and mean dwell time are assessed using the Bonferroni test(p < 0.017);#,Significant value after Bonferroni correction;RD, retinal detachment; HCs, healthy controls;dFC,dynamic functional connectivity;dEC, dynamic effective connectivity;X to Y,from seed region to whole brain;Y to X,from whole brain to seed region.

3.5. SVM classification results

Figure 6A illustrates the SVM classification of dFC in the right V1 and the entire brain.The accuracy and area under the curve (AUC) values were determined to be 0.712 and 0.729, respectively.Figure 6B illustrates the SVM classification of dFC in the left V1 and the entire brain.The accuracy and AUC values were determined to be 0.695 and 0.786, respectively.Figure 6C illustrates the SVM classification of dEC from the right V1 to the whole brain.The accuracy and AUC values were determined to be 0.525 and 0.492, respectively.Figure 6D illustrates the SVM classification of dEC from the whole brain to the right V1.The accuracy and AUC values were determined to be 0.542 and 0.561, respectively.Figure 6E illustrates the SVM classification of dEC from the left V1 to the whole brain.The accuracy and AUC values were determined to be 0.593 and 0.572, respectively.Figure 6F illustrates the SVM classification of dEC from the whole brain to the left V1.The accuracy and AUC values were determined to be 0.458 and 0, respectively.The dFC may be helpful in the clinical diagnosis of RD, according to these findings.

3.6.Verification Analyses

The analysis, utilizing varying sliding-window lengths of 60 TRs (120s) and 80 TRs (160s), consistently supported our primary finding of altered dFC and dEC (In Supplementary Materials, Figure S1 and S2, Table S1 and S2).

This study is the first, to the best of our knowledge, to investigate the dynamic changes in brain functional activity in patients with RD using a combination of FC and GCA approaches. The seed region chosen for analysis was V1, based on previous research indicating cerebral neurohomogeneity dysfunction in the visual pathway of patients with RD. FC analysis revealed abnormal dFC changes between the V1 seed region and various brain regions including occipital, frontal, rectus gyrus, temporal, amygdala, parietal, angular gyrus, supramarginal gyrus, thalamus, and cerebellum. Furthermore, using the GCA method, we identified abnormal dynamic causal connection changes between the V1 seed region and other brain regions, including occipital, calcarine, cuneus, frontal, postcentral gyrus, temporal, parietal, angular gyrus, caudate, thalamus, and cerebellum. Analysis using k-means clustering revealed that RD patients exhibited three predominant states of dFC and three or four states of dEC. As compared to HCs, individuals with RD exhibited significant variations in the NT, F, and MDT. Our SVM model achieved accuracies of 0.712, 0.695, 0.525, 0.542, 0.593, and 0.458, respectively, with corresponding AUC values of 0.729, 0.786, 0.492, 0.561, 0.572, and 0. An AUC range of 0.7-0.9 indicates high accuracy. Therefore, dFC values can be used as sensitive biomarkers to differentiate between patients with HCs and RDs.These findings could contribute to a better understanding of the underlying neural mechanisms responsible for visual impairment in patients with RD.

The occipital lobe is situated posterior to the line connecting the occipital parietal fissure and the anterior occipital notch. The calcarine, on the other hand, is located within the occipital lobe, dividing it into the cuneus gyrus and lingual gyrus. In this study, patients with RD showed increased dFC values in the R-V1 and bilateral MOG. Furthermore, there were increased dEC values from the R-MOG to the R-V1, from the R-CAL to the R-V1, from the L-CUN to the R-V1, from the L-PCUN to the L-V1, as well as a decrease in bidirectional dEC values between the L-V1 and the R-PCUN. Kang et al.²⁹ reported a decrease in the amplitude of low-frequency fluctuation (ALFF) values in the right occipital lobe in patients with RD. Shao et al.⁷ observed decreased functional connection density values in the L-CUN and left occipital lobe among middle-aged patients with RD. Similarly, Huang et al.⁶ found decreased regional homogeneity (ReHo) values in the right occipital lobe and bilateral CUN in patients with RD. Our previous study found increased dynamic ALFF values in the left occipital lobe and R-CAL⁹. The occipital lobe plays a primary role in perceiving and processing visual information, making it crucial in complex visual perception processes. The CUN, which collaborates with V1, is a vital component of the occipital lobe responsible for transmitting visual information to extrastriate cortices³⁰ and also contributing to spatial positioning³¹. In this study, utilizing two analytical methods, we consistently observed higher variability in dFC/dEC in the right V1 and middle occipital gyrus of patients with RD. It is hypothesized that the detached part of the retina in these patients leads to diminished perception of light stimulation, resulting in weakened visual signals received by the occipital lobe. These elevated dFC/dEC levels suggest that people with RD may have a brain compensatory mechanism to offset vision loss.

The frontal lobe serves as the cognitive control center of the brain and plays a crucial role in regulating cognitive function³². The basal surface of the frontal lobe is composed of the rectus gyrus and the orbital gyrus. In this study, patients with RD exhibited increased dFC values between R-V1 and L-SFGmed, as well as R-V1 and R-MFG, while decreased dFC values were observed between L-V1 and R-REC. Additionally, the dEC values showed an increase from R-V1 to R-SFGorb, R-V1 to R-MFGorb, R-V1 to R-MFG, R-V1 to L-MFG, and L-V1 to L-SFG. Kang et al.²⁹ reported a decrease in the ALFF value of the R-MFG in patients with RD, while the ALFF value of the R-SFGorb increased. Shao et al.⁷ observed an increase in the functional connection density values of the L-SFG and L-MFG in middle-aged patients with RD. They proposed that the long-term decline in vision in these patients may impair memory function and stimulate frontal lobe function, which could explain the observed increase in the functional connection density values of L-SFG and L-MFG. Similarly, Huang et al.⁶ found a decrease in the ReHo value of L-MFG in patients with RD, suggesting cognitive impairment in these individuals. Our previous study revealed an increase in the dynamic ALFF values of the bilateral MFG and the right inferior frontal gyrus (IFG) in patients with RD⁹, which aligns with our current research. Patients who experience RD face challenges in adapting to monocular vision due to the sudden loss of vision. When the affected eye no longer transmits visual information to the corresponding brain area, the frontal lobe compensates by assuming the visual processing function.This compensatory response may be an adaptive mechanism to address the visual function defect caused by vision loss in patients with RD.

The parietal lobe, a crucial region in the brain, plays a vital role in spatial visual processing. It is comprised of functional regions such as the angular gyrus, supramarginal gyrus, and postcentral gyrus. This study found significant increases in dFC values between L-V1 and bilateral SPG, R-V1 and L-ANG, as well as R-V1 and R-SMG in RD patients. Additionally, there was a significant increase in dEC values observed from R-V1 to R-IPL, L-V1 to R-IPL, R-V1 to R-ANG, and R-IPL to R-V1. Conversely, a significant decrease in dEC values was found from L-PoCG to L-V1. In a study by Wen et al.³³, an increase in dALFF value was reported in the SPG of patients with active thyroid-associated ophthalmopathy. This finding was postulated to be associated with delayed visuospatial information processing.Wu et al.³⁴ discovered a decrease in FC values of the IPL among patients with asthma, potentially influencing attention and executive function. They also found a decrease in cortical thickness values in the L-IPL and R-SPG among high myopia patients, suggesting structural changes impacting associated cognitive and executive functions³⁵. In our previous study, we observed a decrease in dynamic ALFF value in the R-SPG among patients with RD. As the SPG plays a role in the transmission and integration of visual information, this finding may contribute to the decline in vision experienced by RD patients⁹. The parietal lobe, positioned adjacent to the occipital lobe, is crucial in the visual pathway by transmitting visual information to the frontal lobe^36,37. Given the absence of retinal function and the inability to perceive external visual signals, it can be inferred that the brain region responsible for visual function in RD patients with long-term low vision experiences a prolonged lack or reduced level of stimulation. Consequently, to compensate for the reduced function in this brain area, its activity exhibits an increased level.

The temporal lobe and the amygdala are two important structures in the brain involved in cognitive and emotional processes. In this study, dFC values demonstrated an increase between the R-V1 and the L-STG, while they decreased between the L-V1 and L-STG, as well as between L-V1 and the R-AMYG in RD patients. Additionally, there was an increase in dEC values observed from L-STG to R-V1 and from the L-MTG to R-V1. Shao et al.⁷ observed an increase in the functional connection density of the bilateral inferior temporal gyrus in middle-aged RD patients. This brain region is known to be involved in complex object feature characterization and face perception. The authors postulated that this increased density may represent a compensatory mechanism for the visual decline experienced by middle-aged RD patients. Chen et al.³⁸ found elevated degree centrality (DC) values in the L-MTG of individuals diagnosed with primary angle-closure glaucoma. The researchers speculated that these patients may experience cognitive difficulties. Wen et al.³⁹ also discovered diminished FC between the R-ANG and right superior temporal gyrus (R-STG) in individuals diagnosed with thyroid-associated eye disease. The researchers suggested that these patients might experience cognitive changes. In our previous study, we observed an increase in the dynamic ALFF value of the R-MTG in patients with RD. This may potentially indicate a compensatory mechanism for the reduced language comprehension ability⁹. Language and cognition are closely intertwined and play a fundamental role in human communication and thought processes. The MTG plays a pivotal role as a constituent of the DMN, a network implicated in various cognitive processes such as emotion regulation, self-reflection, and memory⁴⁰. Therefore, our speculation revolves around the hypothesis that long-term visual impairment could contribute to a deterioration in patients' ability to perceive external stimuli, potentially resulting in cognitive problems to some degree. However, further investigation is necessary to elucidate the underlying mechanism in detail.

The thalamus, caudate nucleus, and cerebellum are three intricate brain structures, each with distinct roles and functions. The thalamus, a key structure in the forebrain, serves a critical function in sensory transduction, particularly in the visual system⁴¹. Within the thalamus, the lateral geniculate nucleus receives signals from retinal cells ^42,43. The caudate nucleus, located in the striatum of the basal ganglia, actively participates in the corticothalamic circuit and plays important roles in motor and cognitive function⁴⁴. The cerebellum is crucial for functional interaction with the frontal eye fields, contributing to visuomotor coordination, higher-level cognitive functions, and memory processes^45,46.In our previous research, we found a correlation between high myopia and decreased gray matter volume (GMV) in the R-THA, suggesting a potential contribution of high myopia to thalamic dysfunction⁴⁷. Qi et al.⁴⁸ identified heightened FC between bilateral V1 and bilateral caudate in individuals with diabetic retinopathy, suggesting an augmentation of visuomotor function in these patients. Tong et al.⁴⁹ also observed increased FC between V1 and specific regions of the cerebellum (left cerebellum crus 1 and left cerebellum 10) in individuals with iridocyclitis, suggesting a compensatory response to vision loss.In our study, significant increases in dFC values were observed between R-V1 and R-THA, as well as between R-V1 and cerebellum-vermis-9, in patients with RD. On the other hand, a decreased dFC value was found between L-V1 and R-cerebellum-crus-I. Additionally, significant increases in dEC values were observed from R-V1 to L-CAU and from R-THA to R-V1. Therefore, we speculate that patients with RD may experience difficulties in visual information transmission, visual-motor coordination, and cognition.

In all subjects, we observed the presence of three stable and recurring dFC states, as well as three or four dEC states. Depending on the ROI, there were variations in the NT, F, and MDT. In comparison to the HCs, the RD group exhibited significantly higher frequencies in state 4 of the dEC clustering results from the R-V1 to the whole brain (p = 0.005, t = 2.892). According to these results, people with RD may mostly exhibit state 4 brain activity patterns. MDT, F, and NT are commonly used terms to describe the dynamic temporal characteristics of brain activity, which can be altered in the presence of specific diseases⁵⁰. Furthermore, Li et al.⁵¹propose that state transitions partially reflect the stability of neural activity, while a previous study has demonstrated that brains affected by cognitive dysfunction exhibit lower stability in neural activity⁵². Therefore, we hypothesize that these temporal characteristics of dEC may serve as potential biomarkers of cognitive impairment in RD patients.

Notably,we investigated whether differences in dFC and dEC between individuals with RD and HCs could be utilized as a classifier to differentiate these groups. To address this, we utilized a machine learning approach by employing SVM classifiers based on dFC/dEC values derived from various ROIs.The accuracy of the SVM model to differentiate RD patients and HCs was found to be 0.712, 0.695, 0.525, 0.542, 0.593, and 0.458, respectively. Correspondingly, the AUC were 0.729, 0.786, 0.492, 0.561, 0.572, and 0, respectively.These results suggest that dFC may hold promise as a valuable tool for classifying individuals with RD from the healthy control population.

However, the study has several limitations.Firstly, future research should enlarge the sample size because it is currently too small.Secondly, physiological indicators such as respiratory rate, individual blood oxygen levels, and heart rate were not excluded and could potentially confound spontaneous neuronal activity.Thirdly, the study did not employ multimodal MRI methods to validate the results.

In this study, we employed seed-based FC analysis, GCA, K-means clustering, SVM, and correlation analysis to examine alterations in dynamic function and effective connectivity in the V1 among patients with RD.This study aims to elucidate the neural mechanisms that underlie a visual impairment in patients with RD and to propose that variability in dFC could serve as a valuable index for clinical diagnosis and assessment.

Data Availability Statement

MRI data collection for this project was conducted at Jiangxi Provincial Medical Imaging Clinical Research Center/Clinical Research Center For Medical Imaging In Jiangxi Province (No.20223BCG74001).

Ethics Statement

The studies involving human participants were reviewed and approved by Declaration of Helsinki and was approved by the medical ethics committee of the First Affiliated Hospital of Nanchang university .The research program was approved by the institutional review board of the First Affiliated Hospital of Nanchang Nanchang university. The participants provided their written informed consent to participate in this study.

Author Contributions

JY responsible for writing manuscript; WYY is in charge of proofreading and refining the manuscript's wording.CQ FWW SBL WB HQY contributed to data collection, statistical analyses. JY and WYY designed the protocol and contributed to the MRI analysis.JY WYY and WXR designed the study, oversaw all clinical aspects of study conduct, and manuscript preparation. All authors contributed to the article and approved the submitted version.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Murakami, T. et al. Association between perifoveal hyperfluorescence and serous retinal detachment in diabetic macular edema. Ophthalmology 120, 2596–2603 (2013).
Mattioli, S. et al. Physical exertion (lifting) and retinal detachment among people with myopia. Epidemiology 19, 868–871 (2008).
Shahid, A. et al. Risk Factors Associated With Rhegmatogenous Retinal Detachment. Cureus 14, e23201 (2022).
Beheiri, S. I. et al. Incidence and Risk Factors for Retinal Detachment Following Cataract Surgery at a Tertiary Center in Jeddah, Saudi Arabia. Cureus 15, e33968 (2023).
Etminan, M., Forooghian, F., Brophy, J. M., Bird, S. T. & Maberley, D. Oral fluoroquinolones and the risk of retinal detachment. JAMA 307, 1414–1419 (2012).
Huang, X. et al. Abnormal regional spontaneous neural activity in visual pathway in retinal detachment patients: a resting-state functional MRI study. Neuropsychiatr Dis Treat 13, 2849–2854 (2017).
Shao, Y. et al. Functional connectivity density alterations in middle-age retinal detachment patients. Brain Behav 11, e01783 (2021).
Yang, Y. C. et al. Investigation of Changes in Retinal Detachment-Related Brain Region Activities and Functions Using the Percent Amplitude of Fluctuation Method: A Resting-State Functional Magnetic Resonance Imaging Study. Neuropsychiatr Dis Treat 17, 251–260 (2021).
Ji, Y. et al. Machine learning analysis reveals aberrant dynamic changes in amplitude of low-frequency fluctuations among patients with retinal detachment. Front Neurosci 17, 1227081 (2023).
van den Heuvel, M. P. & Hulshoff Pol, H. E. Exploring the brain network: a review on resting-state fMRI functional connectivity. Eur Neuropsychopharmacol 20, 519–534 (2010).
Zang, Z. X., Yan, C. G., Dong, Z. Y., Huang, J. & Zang, Y. F. Granger causality analysis implementation on MATLAB: a graphic user interface toolkit for fMRI data processing. J Neurosci Methods 203, 418–426 (2012).
Xue, J. et al. Altered Directed Functional Connectivity of the Hippocampus in Mild Cognitive Impairment and Alzheimer's Disease: A Resting-State fMRI Study. Front Aging Neurosci 11, 326 (2019).
Song, Z., Chen, J., Wen, Z. & Zhang, L. Abnormal functional connectivity and effective connectivity between the default mode network and attention networks in patients with alcohol-use disorder. Acta Radiol 62, 251–259 (2021).
Zhou, G. P. et al. Aberrant functional and effective connectivity of the frontostriatal network in unilateral acute tinnitus patients with hearing loss. Brain Imaging Behav 16, 151–160 (2022).
Liu, X. & Duyn, J. H. Time-varying functional network information extracted from brief instances of spontaneous brain activity. Proc Natl Acad Sci U S A 110, 4392–4397 (2013).
Zhang, Y. & Wu, L. Classification of fruits using computer vision and a multiclass support vector machine. Sensors (Basel) 12, 12489–12505 (2012).
Zhang, Y., Dong, Z., Phillips, P., Wang, S., Ji, G., Yang, J. & Yuan, T. F. Detection of subjects and brain regions related to Alzheimer's disease using 3D MRI scans based on eigenbrain and machine learning. Front Comput Neurosci 9, 66 (2015).
Meng, X., Wu, Y., Liang, Y., Zhang, D., Xu, Z., Yang, X. & Meng, L. A Triple-Network Dynamic Connection Study in Alzheimer's Disease. Front Psychiatry 13, 862958 (2022).
Jia, X. Z. et al. RESTplus: an improved toolkit for resting-state functional magnetic resonance imaging data processing. Sci Bull (Beijing) 64, 953–954 (2019).
Friston, K. J., Williams, S., Howard, R., Frackowiak, R. S. & Turner, R. Movement-related effects in fMRI time-series. Magn Reson Med 35, 346–355 (1996).
Yan, C. G. et al. A comprehensive assessment of regional variation in the impact of head micromovements on functional connectomics. Neuroimage 76, 183–201 (2013).
Ji, Y. et al. Exploration of static functional connectivity and dynamic functional connectivity alterations in the primary visual cortex among patients with high myopia via seed-based functional connectivity analysis. Front Neurosci 17, 1126262 (2023).
Liao, W., Wu, G. R., Xu, Q., Ji, G. J., Zhang, Z., Zang, Y. F. & Lu, G. DynamicBC: a MATLAB toolbox for dynamic brain connectome analysis. Brain Connect 4, 780–790 (2014).
Glover, G. H. Overview of functional magnetic resonance imaging. Neurosurg Clin N Am 22, 133–139, vii (2011).
Wu, G. R., Colenbier, N., Van Den Bossche, S., Clauw, K., Johri, A., Tandon, M. & Marinazzo, D. rsHRF: A toolbox for resting-state HRF estimation and deconvolution. Neuroimage 244, 118591 (2021).
Wu, G. R., Liao, W., Stramaglia, S., Ding, J. R., Chen, H. & Marinazzo, D. A blind deconvolution approach to recover effective connectivity brain networks from resting state fMRI data. Med Image Anal 17, 365–374 (2013).
Allen, E. A., Damaraju, E., Plis, S. M., Erhardt, E. B., Eichele, T. & Calhoun, V. D. Tracking whole-brain connectivity dynamics in the resting state. Cereb Cortex 24, 663–676 (2014).
Pereira, F., Mitchell, T. & Botvinick, M. Machine learning classifiers and fMRI: a tutorial overview. Neuroimage 45, S199-209 (2009).
Kang, H. H. et al. Measuring abnormal intrinsic brain activities in patients with retinal detachment using amplitude of low-frequency fluctuation: a resting-state fMRI study. Int J Neurosci 129, 681–686 (2019).
Vanni, S., Tanskanen, T., Seppä, M., Uutela, K. & Hari, R. Coinciding early activation of the human primary visual cortex and anteromedial cuneus. Proc Natl Acad Sci U S A 98, 2776–2780 (2001).
Rao, H., Zhou, T., Zhuo, Y., Fan, S. & Chen, L. Spatiotemporal activation of the two visual pathways in form discrimination and spatial location: a brain mapping study. Hum Brain Mapp 18, 79–89 (2003).
Wong, K., Xu, J., Chen, C., Ghista, D. & Zhao, H. Functional magnetic resonance imaging providing the brain effect mechanism of acupuncture and moxibustion treatment for depression. Front Neurol 14, 1151421 (2023).
Wen, Z., Kang, Y., Zhang, Y., Yang, H., Zhao, Y., Huang, X. & Xie, B. Disrupted dynamic amplitude of low-frequency fluctuations in patients with active thyroid-associated ophthalmopathy. Front Cell Dev Biol 11, 1174688 (2023).
Wu, Y. J. et al. Impaired Interhemispheric Synchrony in Bronchial Asthma. Int J Gen Med 14, 10315–10325 (2021).
Wu, Y. J. et al. Evidence of cortical thickness reduction and disconnection in high myopia. Sci Rep 10, 16239 (2020).
Ioannides, A. A., Liu, L., Poghosyan, V., Saridis, G. A., Gjedde, A., Ptito, M. & Kupers, R. MEG reveals a fast pathway from somatosensory cortex to occipital areas via posterior parietal cortex in a blind subject. Front Hum Neurosci 7, 429 (2013).
Caminiti, R., Ferraina, S. & Johnson, P. B. The sources of visual information to the primate frontal lobe: a novel role for the superior parietal lobule. Cereb Cortex 6, 319–328 (1996).
Chen, R. B., Zhong, Y. L., Liu, H. & Huang, X. Machine learning analysis reveals abnormal functional network hubs in the primary angle-closure glaucoma patients. Front Hum Neurosci 16, 935213 (2022).
Wen, Z., Wan, X., Qi, C. X. & Huang, X. Local-to-Remote Brain Functional Connectivity in Patients with Thyroid-Associated Ophthalmopathy and Assessment of Its Predictive Value Using Machine Learning. Int J Gen Med 15, 4273–4283 (2022).
Tong, Y., Huang, X., Qi, C. X. & Shen, Y. Disrupted Neural Activity in Individuals With Iridocyclitis Using Regional Homogeneity: A Resting-State Functional Magnetic Resonance Imaging Study. Front Neurol 12, 609929 (2021).
Usrey, W. M. & Alitto, H. J. Visual Functions of the Thalamus. Annu Rev Vis Sci 1, 351–371 (2015).
Diao, Y. et al. Reciprocal Connections Between Cortex and Thalamus Contribute to Retinal Axon Targeting to Dorsal Lateral Geniculate Nucleus. Cereb Cortex 28, 1168–1182 (2018).
Williams, A. L., Reese, B. E. & Jeffery, G. Role of retinal afferents in regulating growth and shape of the lateral geniculate nucleus. J Comp Neurol 445, 269–277 (2002).
Persson, K., Bohbot, V. D., Bogdanovic, N., Selbaek, G., Braekhus, A. & Engedal, K. Finding of increased caudate nucleus in patients with Alzheimer's disease. Acta Neurol Scand 137, 224–232 (2018).
Desmond, J. E. & Fiez, J. A. Neuroimaging studies of the cerebellum: language, learning and memory. Trends Cogn Sci 2, 355–362 (1998).
Stoodley, C. J. & Schmahmann, J. D. Evidence for topographic organization in the cerebellum of motor control versus cognitive and affective processing. Cortex 46, 831–844 (2010).
Huang, X. et al. Altered whole-brain gray matter volume in high myopia patients: a voxel-based morphometry study. Neuroreport 29, 760–767 (2018).
Qi, C. X., Huang, X., Tong, Y. & Shen, Y. Altered Functional Connectivity Strength of Primary Visual Cortex in Subjects with Diabetic Retinopathy. Diabetes Metab Syndr Obes 14, 3209–3219 (2021).
Tong, Y., Huang, X., Qi, C. X. & Shen, Y. Altered Functional Connectivity of the Primary Visual Cortex in Patients With Iridocyclitis and Assessment of Its Predictive Value Using Machine Learning. Front Immunol 12, 660554 (2021).
Xu, L. et al. Dynamic alterations of spontaneous neural activity in post-stroke aphasia: a resting-state functional magnetic resonance imaging study. Front Neurosci 17, 1177930 (2023).
Li, J. et al. High transition frequencies of dynamic functional connectivity states in the creative brain. Sci Rep 7, 46072 (2017).
Hellyer, P. J., Scott, G., Shanahan, M., Sharp, D. J. & Leech, R. Cognitive Flexibility through Metastable Neural Dynamics Is Disrupted by Damage to the Structural Connectome. J Neurosci 35, 9050–9063 (2015).

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Aberrant dynamic functional and effective connectivity changes of the primary visual cortex in patients with retinal detachment via machine learning

Status:

Version 1

Abstract

Figures

1. Introduction

2. Participants and methods

2.1. Participants

2.2. fMRI Data Acquisition

2.3. fMRI Data Preprocessing

2.4. Seed-based dFC analysis

2.5. Blind Deconvolution Procedure

2.6. Seed-based dEC analysis

2.7. Statistical analysis

2.8. Clustering analysis

2.9. Support vector machine analysis

2.10.Validation analysis

3. Results

3.1. Demographic characteristics

4. Discussion

5. Limitations

6. Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1