Alterations of static and dynamic functional network connectivity in acute ischemic brainstem stroke

Abstract

Background

Prior studies have shown abnormal brain functional network changes in patients with acute ischemic stroke. However, the alterations of dynamic functional network connectivity (FNC) in brainstem strokes have not been elucidated.

Purpose

To assess alterations of static and dynamic FNCs and determine the relationships between these and upper limb movement performance in patients with acute brainstem ischemic stroke.

Material and Methods

In total, 50 patients with acute brainstem ischemic stroke and 50 age- and sex-matched healthy controls were enrolled in the present study and underwent resting-state functional magnetic resonance imaging (rs-fMRI). Independent component analysis was conducted to assess static and dynamic FNC patterns based on seven resting-state networks, namely, the default mode network (DMN), executive control network (ECN), attention network (AN), somatomotor network (SMN), visual network (VN), auditory network (AUN), and cerebellum network (CN).

Results

Compared with controls, patients with acute brainstem ischemic stroke exhibited wide aberrations of static FNC, including increased FNC in DMN–ECN, DMN–VN, ECN–VN, ECN–AN and AN–AUN pairs. Patients with acute brainstem ischemic stroke showed aberrant dynamic FNC in State 1, involving increased FNC aberrance in the DMN with AN, DMN with ECN, and reduced FNC in SMN–VN pairs. In State 5, patients with acute brainstem ischemic stroke showed increased FNC in DMN–VN and AN–AUN, and decreased FNC in AN–SMN pairs.

Conclusion

This study suggests that static and dynamic FNC impairment and aberrant connections exist in acute brainstem ischemic stroke, which expands what is known regarding the relationship between stroke and FNC from static and dynamic perspectives.

Keywords

Brainstem ischemic stroke resting-state network static functional network connectivity dynamic functional network connectivity functional magnetic resonance imaging

Introduction

Acute ischemic stroke (AIS) is a frequently occurring disease, and the incidence is showing a younger trend (1,2). More than one-third of patients with AIS have lateral movement and cognitive disorders (3). Brainstem ischemic stroke accounted for 21.9% (4), with a rapid onset and greater risk than ischemic stroke in other parts of the brain. The evaluation of early cognitive and movement disorders in patients with acute brainstem ischemic stroke is extremely important.

In recent years, resting-state functional magnetic resonance imaging (rs-fMRI), which measures the features of low-frequency blood oxygenation level-dependent (BOLD) fluctuations (5,6), has grown to become a reliable method for exploring human brain functional connectivity (FC) (7 –9). Functional neuroimaging has uncovered neural mechanisms involved in post-stroke plasticity and reorganization (10 –12). A highly consistent finding encountered in stroke is the decline of interhemispheric connectivity between the primary sensorimotor cortices, which develops in the first weeks after stroke and returns to levels observed in healthy individuals in parallel with behavioral recovery (13 –15). Compared with healthy controls (HCs), patients with AIS showed significantly decreased functional connectivity in the right medial prefrontal cortex (mPFC) and right precuneus within the anterior and posterior DMN, respectively (16). Static functional network connectivity (sFNC) can be used to assess temporal correlations between two brain regions throughout the fMRI acquisition; however, applicability of this is limited by oversimplified analysis that does not include temporal dynamics (17). Dynamic functional network connectivity (dFNC) analyses can be used as distinct “connectivity states” of the brain by summarizing reoccurring large-scale patterns of connectivity and transition trajectories between them. These dynamic measures allow for a more sophisticated assessment of the spontaneous fluctuating nature of neural signals compared to static ones and are increasingly being used as novel biomarkers of disease (18,19). dFNC and sFNC should be employed together in one analysis, as the former can be considered a dynamically changing ensemble. In addition, it is difficult to compare the mapping from the observed dynamic coactivation to the patterns observed in sFNC, so the two techniques may yield different results and capture complementary information as neuroimaging biomarkers (20).

Therefore, the aim of the present study was to explore the underlying neural mechanisms of upper extremity dyskinesia after acute brainstem ischemic stroke by combining dFNC and sFNC based on the rs-fMRI approach. We hypothesized that patients with acute brainstem ischemic stroke would exhibit disrupted sFNC and dFNC patterns and that these abnormalities were associated with upper extremity motor dysfunction.

Material and Methods

Participants and clinical data

All patients in this study provided written informed consent before examination. The study was approved by the local ethics committee of Taizhou People's Hospital. A total of 50 patients (23 men, 27 women; age range = 44–80 years) with acute brainstem ischemic stroke were recruited from the emergency department between January 2019 and January 2022. MRI examination was performed 24 h after the patient was admitted to the emergency department. The inclusion criteria for patients with acute brainstem ischemic stroke were as follows: (i) aged 40–80 years; (ii) initial emergency room evaluation of acute brainstem ischemic stroke (accompanied by physical weakness, language impairment, and other symptoms of acute cerebral stroke); (iii) an initial Glasgow Coma Score (GCS) of 12–15 at the emergency department; (iv) a single unilateral lesion located in the pons as identified by MRI examination; and (v) computed tomography scan as clinical evaluation to rule out cerebral hemorrhage. The exclusion criteria were as follows: (i) large-scale infarct lesions; (ii) history of sedative use in hospitals or emergency rooms; (iii) history of drug or alcohol misuse; (iv) history of a preceding acute cerebral stroke; and (v) MRI contraindications. In addition, 50 age- and sex-matched healthy volunteers (23 men, 27 women) were recruited as the HC group. The same exclusion criteria were adopted as the patient group. To assess the stroke severity of each patient, we used the National Institute of Health Stroke Scale (NIHSS) and Fugl-Meyer assessment (FMA). FMA was performed to assess upper extremity motor disorder. This test consisted of wrist, elbow, shoulder, forearm, hand movement, and coordination (21).

Data acquisition

All MR scans were performed with a 3.0-T MRI scanner (Skyra, Siemens, Enlargen, Germany) with an eight-channel receiver array head coil. During scanning, participants were asked to lie quietly and keep still as much as possible with their eyes closed but avoid sleep or special thinking activities. Foam padding and earplugs were used to minimize head movement and the disturbance of scanning noise. The sequence parameters of the rs-fMRI, high-resolution T1-weighted (T1W) images, and susceptibility-weighted images are shown in the Supplemental Materials.

Sequence parameters

The rs-fMRI data were acquired by gradient echo-planer imaging sequences with the following parameters: repetition time (TR)/echo time (TE) = 2000/25 ms; slices = 36; thickness = 4 mm; gap =0 mm; field of view (FOV) = 240 × 240 mm; acquisition matrix = 64 × 64; and flip angle (FA) = 90°. The fMRI sequence scan took 8 min 8 s. The high-resolution T1W images were obtained with a T1W 3D spoiled gradient-echo sequence as follows: TR/TE = 1900/2.48 ms; thickness = 1 mm; slices = 170; FA = 8°; FOV = 256 × 256 mm; gap = 0 mm; and acquisition matrix = 256 × 256. The structural sequence scan took 5 min 26 s. In addition, 3D gradient echo susceptibility-weighted imaging sequences (TR/ TE = 22/34 ms; FA = 20°; matrix = 276 × 319; slice thickness = 1 mm; and FOV = 220 × 220 mm) were also implemented to help detect hemorrhagic or other lesions.

Data preprocessing

GRETNA (Graph Theoretical Network Analysis) was applied to preprocess the rs-fMRI data for further analysis with the following stages (22). The first 10 volumes of each time series were removed to allow for participant adaptation to the scanning environment. The odd and middle slices were the starting and reference slices, respectively. Participant data, including demonstrating head movement >2.0 mm translation or >2.0° rotation, were excluded from the analysis. The rest of the dataset was spatially normalized to a template from the Montreal Neurological Institute (resampled voxel size = 3 × 3 × 3 mm), followed by a 6-mm spatial sequence with a Gaussian smoothing kernel.

Independent component analysis and resting-state network selection

Resting-state networks (RSNs) were selected with the group independent component analysis (ICA) software of fMRI toolbox software. ICA analysis was performed in three phases: (i) data reduction; (ii) application of the ICA algorithm; and (iii) back reconstruction for each individual. The number of independent components (ICs) was determined using the minimum description length (MDL) criteria (23). In phase one, the computational complexity was decreased with principal component analysis (PCA), and the remaining reduction step was achieved again using PCA in light of a selected number of ICs. In phase two, we ran the proper ICA through the infomax algorithm. Finally, group ICA (GICA) type back reconstruction was performed on single-subject individual time course and spatial maps, and the results were converted to z scores for display.

Among the 32 ICs, 20 artifactual components were discarded, including edges of the brain or ventricles, the peak cluster locations in white matter showing minimal overlap with gray matter, and a reduced low-frequency/high-frequency activity ratio (17). Finally, 12 ICs were obtained and categorized into seven RSNs: default mode network (DMN); executive control network (ECN); attention network (AN); somatomotor network (SMN); visual network (VN); auditory network (AUN); and cerebellum network (CN). These seven RSNs have been commonly reported in previous rs-fMRI studies (24,25). The networks were determined using the spatial correlation similarity metric between ICA components and a template of the desired RSNs (26).

sFNC analysis

The individual-level time courses of the identified RSNs were obtained by the spatiotemporal double regression method after ICA. Then, the relationship between different RSN time courses was analyzed. A time-domain bandpass filter (bandpass = 0.00–0.25 Hz) was used to decrease the influence of low-frequency drift and high-frequency physiological noise on the time process during the analysis. Second, correlations between any two RSN time courses of each subject were calculated. Then, the FNC known as temporal correlation was acquired by calculating the Pearson correlation coefficient of the time courses of selected RSNs and generating the matrix of 12 × 12 (RSNs) × 100 (participants). In a general linear model, age and sex were finally used as covariates to analyze which RSN pairs were significantly different between controls and patients. The significance threshold was P <0.05, corrected for multiple comparisons using FDR

dFNC analysis

The temporal dFNC toolbox in the GIFT package from MATLAB was used to perform dFNC analysis. The dFNC between ICA time courses was computed using a sliding time-window approach. The dFNC sliding window size was set to 15 TRs (30 s) of a rectangular window convolved with a Gaussian (= 3 TRs) (27). K-means clustering using squared Euclidean distance, 500 iterations, and 150 replicate dFNC windows (28) was used to assess the recurrence of FNC patterns. The number of clusters was determined by the value of k, in the range of 2–10, which was defined as the ratio of intra- to inter-cluster distances. Based on the elbow criterion (25), the optimal number of clusters is inferred as a point on the elbow of the curve, representing the beginning of a small reduction in the within-cluster variability. According to the elbow criterion results, the optimal number of clusters was determined to be 5 (k = 5). All the dFNC matrices for each subject were categorized into one of the five clusters based on the similarity with the cluster centroid. The following dFNC indices were computed for each subject (25): (i) mean dwell time, computed as the average time wherein each subject was at a specific state; (ii) occurrence frequency of state, indicating the percentage of each state in a group of five states; and (iii) transition number, indicating the number of times the subject switched between states.

Statistical analysis

An independent t test for continuous variables and a chi-square test for proportions were used to assess demographic and clinical characteristics between patients with stroke and HCs using SPSS 25.0 software (IBM Corp., Armonk, NY, USA). The significance threshold was set as P <0.05. Shapiro‒Wilk tests were used to evaluate data normality, and P >0.05 indicated a normal distribution of data. For RSN and FNC analysis, group comparisons between patients with stroke and HCs were performed using two-sample t tests. Age and sex were used as covariates. Correction for multiple comparisons was carried out using the false discovery rate (FDR) with the Benjamini–Hochberg (BH) procedure (P_FDR <0.05). Two-sample t tests were performed on each of the 66 mean dFNC correlations from each of the five states. FDR (P_FDR <0.05) was used to correct group difference results for multiple comparisons. Data from patients and HCs were separated for each state and used for further analysis. dFNC indices were compared between the two groups by using the non-parametric Mann–Whitney U test (P <0.05). Correction for multiple comparisons was carried out using Bonferroni calibration for the five states and different indices (P_Bonferroni <0.05).

dFNC values and FMA scores were evaluated using Spearman's correlation coefficients among FNC z scores, dFNC values (mean dwell time and occurrence frequency of state), and FMA scores. P <0.05 was considered significant.

Results

Demographics data

A total of 50 patients and 50 HCs were finally enrolled. None of the participants were excluded due to image distortion or head motion. All patients had an initial NIHSS score of 3.64 ± 1.57 at the emergency department (Table 1). No significant differences in age, sex, or level of education observed between the two groups (all P >0.05).

Table 1.

Demographic characteristics in stroke patients and healthy controls.

Characteristics	Stroke (n = 50)	Controls (n = 50)	P value
Age (years)	62.08 ± 7.71	60.72 ± 4.24	0.278
Sex (M/F)	23/27	23/27	1.000
Level of education (years)	10.72 ± 1.96	10.70 ± 1.61	0.956
NIHSS	3.64 ± 1.57	--	--
FMA	56.66 ± 7.22	--	--

Values are given as mean ± SD.

FMA, Fugl-Meyer assessment; NIHSS, National Institute of Health stroke scale.

RSN results

The seven networks with 12 components shown in Fig. 1 are as follows: (i) the default mode network (DMN) (IC9 + 11 + 12 + 14) typically included the posterior cingulate cortex (PCC), medial prefrontal cortex (MPFC), inferior parietal, precuneus, and bilateral angular gyrus nodes. The executive control network (ECN) (IC20 + 22) included the left lateral frontoparietal network (LFPN) and the right lateral frontoparietal network (RFPN). The LFPN was mainly focused on the left middle frontal gyrus, superior parietal lobule, inferior parietal lobule, and angular gyrus. The spatial distribution of RFPN was similar to that of LFPN. The attention network (AN) (IC7) included the frontal eye field, intraparietal sulcus, superior parietal lobe, inferior temporal gyrus, middle frontal gyrus, and inferior parietal gyrus. The visual network (VN) (IC2 + 10) included the primary visual cortex (the bilateral calcarine sulci and medial extrastriate regions [e.g. the lingual gyrus and cuneus]) and the occipital part of the fusiform gyrus and extravisual cortical regions (the occipital pole, the lateral occipital cortex). The auditory network (AUN) (IC21) mainly included the bilateral intraparietal sulcus, middle temporal lobe, and frontal eye field. The sensorimotor network (SMN) (IC6) comprised the SMN1, which included the paracentral lobule, the supplementary motor area (SMA), and the pre- and postcentral gyri; the SMN2 was mainly focused on the bilateral primary somatosensory cortex, including the postcentral gyri areas and precentral area. The cerebellum network (CN) (IC15) had spatial patterns that primarily encompassed the anterior lobe, posterior lobe, and declive of the cerebellum.

sFNC and dFNC results

Compared with the HC group, the acute brainstem ischemic stroke group exhibited significantly reduced sFNC in the DMN (IC12)–DMN (IC14) pairs and significantly increased sFNC in the DMN (IC14)–ECN (IC20), DMN (IC12)–VN (IC2), ECN (IC22)–VN (IC2), ECN (IC20)–AN (IC7) and AN (IC7)–AUN (IC21) pairs (Fig. 2a).

Fig. 3 illustrates the dFNC states characterized by the cluster centroids. State 1 had the greatest frequency (31%). The total number of individuals in each state varied because not all patients visited all states. The participant distribution among dFNC states is shown in Fig. 3. The majority of the individuals visited States 1 (53/87) and 5 (66/87), which are also the states with the largest occupancy rates. Only 31 participants visited State 2, 14 visited State 3, and 39 visited State 4 (Fig. 4a). Significant differences between the two groups were found in State 1 (Fig. 4b). Increased FNC aberrance was found in the DMN with AN, DMN with ECN, and reduced FNC in SMN–VN pairs. In State 5 (Fig. 4c), patients with acute brainstem ischemic stroke showed increased FNC in DMN–VN and AN–AUN, and decreased FNC in SMN–AN pairs.

The patients with acute brainstem ischemic stroke had a significantly longer mean dwell time in States 3 and 5 (P_Bonferroni = 0.030, 0.000, separately) and a significantly shorter mean dwell time in State 1 (P_Bonferroni = 0.015) (Table 2) than the HCs. The patients with acute brainstem ischemic stroke also had a significantly higher occurrence frequency of State 5 (P_Bonferroni = 0.000) and a lower occurrence frequency of State 1 (P_Bonferroni = 0.000) (Table 3) than the HCs. However, no significant difference in the transition number of states was found between the patients and HCs (4.37 ± 2.85 vs. 3.28 ± 2.01; P = 0.260).

Table 2.

Mean dwell time between stroke patients and healthy controls.

State	Stroke (n = 50)	Controls (n = 50)	P_Bonferroni value
1	20.28 ± 32.05	50.98 ± 57.09	0.015*
2	17.23 ± 32.14	13.40 ± 21.02	2.590
3	15.05 ± 29.53	1.87 ± 8.85	0.030*
4	15.02 ± 30.13	21.40 ± 24.89	1.440
5	53.63 ± 57.08	17.93 ± 22.31	0.000*

Values are given as mean ± SD. The P_Bonferroni values are obtained by using the Bonferroni calibration. N for Bonferroni corrections = 5.

*P_Bonferroni <0.05 indicates a significant difference.

Table 3.

Occurrence frequency of states between stroke patients and controls.

State	Stroke (n = 50)	Controls (n = 50)	P_Bonferroni value
1	0.17 ± 0.26	0.47 ± 0.39	0.000*
2	0.17 ± 0.29	0.14 ± 0.22	3.080
3	0.09 ± 0.18	0.02 ± 0.09	0.055
4	0.13 ± 0.24	0.22 ± 0.25	0.650
5	0.43 ± 0.34	0.16 ± 0.18	0.000*

Values are given as mean ± SD. The P_Bonferroni values are obtained by using the Bonferroni calibration. N for Bonferroni corrections = 5.

*P_Bonferroni <0.05 indicates significant difference.

Relationship among sFNC, dFNC, and upper extremity function

Increased sFNC in the ECN–VN pair was negatively correlated with the FMA score. (rho = –0.377; P = 0.009) (Fig. 2b). However, no significant correlations among mean dwell time, occurrence frequency of states, and FMA scores were found in patients with acute brainstem ischemic stroke (State 1: r = 0.105, P = 0.489; r = 0.083, P = 0.582; State 5: r = 0.122, P = 0.418; r = 0.016, P = 0.916, respectively).

Discussion

The DMN is in charge of both primary perception and advanced cognition, as well as the integration between them (29). Our previous study found that the FC of the right frontal inferior orbital gyrus (R_FIO) and left temporal middle gyrus (L_TM) in the DMN of patients with acute brainstem ischemic stroke was reduced (30), and the reduction in the DMN (IC12)–DMN (IC14) in this study is consistent with this previous study, suggesting that dysfunction in the DMN can lead to cognitive impairment in stroke patients. The ECN is involved in numerous advanced cognitive tasks and plays a pivotal role in adaptive cognitive control (31). Furthermore, we observed enhanced network connectivity between the DMN–ECN and the DMN–VN in both static and dynamic brain networks, a result that could be explained by the recruitment or compensatory reallocation of cognitive resources.

The AN comprises the bilateral ventrolateral prefrontal cortex and bilateral temporal parietal junction (32,33) and is responsible for interpreting unattended or unexpected stimuli and triggering attention shifts (34). Furthermore, we observed increased sFNC in the ECN–VN pair, which was negatively correlated with the FMA score. The VN mostly encompasses the bilateral occipital lobe. Visual field deficits after stroke have been demonstrated, and enhanced resting-state FC associated with the VN can be observed one month after stroke, predicting recovery of visual deficits (35).

The SMN plays a role in disease-related functional alterations and executing responses to external stimuli and internally generated movements. Extensive changes in connectivity in the functional network occur immediately after acute stroke and are important for recovery (36,37). Rs-fMRI studies have demonstrated a significant reduction in FC between the ipsilateral and contralateral primary sensorimotor cortex in AIS (14,38,39). Decreased FC in the ipsilateral primary motor cortex (M1) was also found in other brain regions, such as the bilateral secondary somatosensory cortex, bilateral SMA, contralateral premotor cortex, and contralateral posterior parietal cortex (12). However, aberrant SMN connectivity was not correlated with any upper extremity movement performance in this study. This finding is possibly due to different methods of ICA analysis and the small sample size.

The AUN contains the lateral superior temporal gyrus, Heschl's gyri, planum temporale, and posterior insula (40,41). These brain areas play roles in cognitive tasks such as language processing and speech (42). The patients with acute brainstem ischemic stroke showed increased sFNC and dFNC between the AN and the AUN. However, the upper extremity motor function of all patients did not correlate with the mean dwell time or frequency of occurrence in each state, nor were the values of dynamic FC, which can be attributed to the small sample size or the excessive states obtained in this study. These findings support our hypothesis and are particularly important for understanding the pathophysiology of stroke, as motor reorganization is a mechanism of cortical damage following subcortical stroke that disrupts the relevant networks supporting motor behavior.

The present study has some limitations. First, this study has a limited sample size and only observed the brain FNC in the acute phase of brainstem stroke. The sample size needs to be expanded to conduct more longitudinal studies on the functional connections between networks during the recovery period of brainstem stroke. Second, our research focused more on the relationship between brain network connections and upper limb dysfunction and lacked an assessment of the connection with cognitive impairment. Finally, although ICA was used to reveal the advantages of unconstrained brain connections, ICA cannot reveal the directionality of projections between networks. Further research is needed to evaluate the direction of each network in interacting with other networks in patients with acute ischemic brainstem stroke.

In conclusion, compared with HCs, patients with acute ischemic brainstem stroke showed sFNC and dFNC aberrances, some of which were associated with upper extremity movement performance. The present findings could help reinforce the understanding of the neural mechanisms and movement disorders which occur after acute ischemic brainstem stroke.

Supplemental Material

sj-docx-1-acr-10.1177_02841851221127271 - Supplemental material for Alterations of static and dynamic functional network connectivity in acute ischemic brainstem stroke

Supplemental material, sj-docx-1-acr-10.1177_02841851221127271 for Alterations of static and dynamic functional network connectivity in acute ischemic brainstem stroke by Jian Zhang and Yi Chang in Acta Radiologica

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Yi Chang

Supplemental material

Supplemental material for this article is available online.

References

Collaborators GBDS. Global, regional, and national burden of stroke and its risk factors, 1990–2019: A systematic analysis for the global burden of disease study 2019. Lancet Neurol 2021;20:795–820.

Owolabi

Thrift

Mahal

, et al. Primary stroke prevention worldwide: Translating evidence into action. Lancet Public Health 2022;7:e74–e85.

Pohjasvaara

Vataja

Leppavuori

, et al. Cognitive functions and depression as predictors of poor outcome 15 months after stroke. Cerebrovasc Dis 2002;14:228–233.

Mortality

GBD

, Causes of death C. Global, regional, and national life expectancy, all-cause mortality, and cause-specific mortality for 249 causes of death, 1980–2015: A systematic analysis for the global burden of disease study 2015. Lancet 2016;388:1459–1544.

Hampson

Peterson

Skudlarski

, et al. Detection of functional connectivity using temporal correlations in MR images. Hum Brain Mapp 2002;15:247–262.

Damoiseaux

Rombouts

Barkhof

, et al. Consistent resting-state networks across healthy subjects. Proc Natl Acad Sci U S A 2006;103:13848–13853.

Biswal

Mennes

Zuo

, et al. Toward discovery science of human brain function. Proc Natl Acad Sci U S A 2010;107:4734–4739.

Guerra-Carrillo

Mackey

Bunge

. Resting-state fMRI: A window into human brain plasticity. Neuroscientist 2014;20:522–533.

Buckner

Krienen

Yeo

. Opportunities and limitations of intrinsic functional connectivity MRI. Nat Neurosci 2013;16:832–837.

10.

Grefkes

Fink

. Reorganization of cerebral networks after stroke: New insights from neuroimaging with connectivity approaches. Brain 2011;134:1264–1276.

11.

Grefkes

Fink

. Connectivity-based approaches in stroke and recovery of function. Lancet Neurol 2014;13:206–216.

12.

Carter

Astafiev

Lang

, et al. Resting interhemispheric functional magnetic resonance imaging connectivity predicts performance after stroke. Ann Neurol 2010;67:365–375.

13.

van Meer

van der Marel

Wang

, et al. Recovery of sensorimotor function after experimental stroke correlates with restoration of resting-state interhemispheric functional connectivity. J Neurosci 2010;30:3964–3972.

14.

Park

Chang

Ohn

, et al. Longitudinal changes of resting-state functional connectivity during motor recovery after stroke. Stroke 2011;42:1357–1362.

15.

Volz

Rehme

Michely

, et al. Shaping early reorganization of neural networks promotes motor function after stroke. Cereb Cortex 2016;26:2882–2894.

16.

Jiang

Geng

Chen

, et al. Decreased functional connectivity within the default-mode network in acute brainstem ischemic stroke. Eur J Radiol 2018;105:221–226.

17.

Allen

Damaraju

Plis

, et al. Tracking whole-brain connectivity dynamics in the resting state. Cereb Cortex 2014;24:663–676.

18.

Preti

Bolton

Van De Ville

. The dynamic functional connectome: State-of-the-art and perspectives. Neuroimage 2017;160:41–54.

19.

Lurie

Kessler

Bassett

, et al. Questions and controversies in the study of time-varying functional connectivity in resting fMRI. Netw Neurosci 2020;4:30–69.

20.

Hutchison

Womelsdorf

Allen

, et al. Dynamic functional connectivity: Promise, issues, and interpretations. Neuroimage 2013;80:360–378.

21.

Platz

Eickhof

van Kaick

, et al. Impairment-oriented training or bobath therapy for severe arm paresis after stroke: A single-blind, multicentre randomized controlled trial. Clin Rehabil 2005;19:714–724.

22.

Wang

Xia

, et al. GRETNA: A graph theoretical network analysis toolbox for imaging connectomics. Front Hum Neurosci 2015;9:386.

23.

Adali

Calhoun

. Estimating the number of independent components for functional magnetic resonance imaging data. Hum Brain Mapp 2007;28:1251–1266.

24.

Wang

Cai

Sun

, et al. Large-Scale internetwork functional connectivity mediates the relationship between serum triglyceride and working memory in young adulthood. Neural Plast 2020;2020:8894868.

25.

Jiang

Shi

Yang

, et al. Aberrant dynamic functional network connectivity in cirrhotic patients without overt hepatic encephalopathy. Eur J Radiol 2020;132:109324.

26.

Mayer

Ling

Allen

, et al. Static and dynamic intrinsic connectivity following mild traumatic brain injury. J Neurotrauma 2015;32:1046–1055.

27.

Pearlson

, et al. Interaction among subsystems within default mode network diminished in schizophrenia patients: A dynamic connectivity approach. Schizophr Res 2016;170:55–65.

28.

Malhi

Das

Outhred

, et al. Resting-state neural network disturbances that underpin the emergence of emotional symptoms in adolescent girls: Resting-state fMRI study. Br J Psychiatry 2019;215:545–551.

29.

Adamaszek

D’Agata

Ferrucci

, et al. Consensus paper: Cerebellum and emotion. Cerebellum 2017;16:552–576.

30.

Geng

Zhang

Shang

, et al. Reduced functional network connectivity is associated with upper limb dysfunction in acute ischemic brainstem stroke. Brain Imaging Behav 2022;16:802–810.

31.

McHugh

Yang

, et al. Executive control network connectivity strength protects against relapse to cocaine use. Addict Biol 2017;22:1790–1801.

32.

Liu

Zhang

, et al. Ventral attention-network effective connectivity predicts individual differences in adolescent depression. J Affect Disord 2019;252:55–59.

33.

Corbetta

Patel

Shulman

. The reorienting system of the human brain: From environment to theory of mind. Neuron 2008;58:306–324.

34.

Vossel

Geng

Fink

. Dorsal and ventral attention systems: Distinct neural circuits but collaborative roles. Neuroscientist 2014;20:150–159.

35.

Kim

Cho

Kim

, et al. Early functional connectivity predicts recovery from visual field defects after stroke. J Stroke 2019;21:207–216.

36.

Grefkes

Ward

. Cortical reorganization after stroke: How much and how functional? Neuroscientist 2014;20:56–70.

37.

Carter

Shulman

Corbetta

. Why use a connectivity-based approach to study stroke and recovery of function? Neuroimage 2012;62:2271–2280.

38.

Wang

Chen

, et al. Dynamic functional reorganization of the motor execution network after stroke. Brain 2010;133:1224–1238.

39.

Golestani

Tymchuk

Demchuk

, et al. Longitudinal evaluation of resting-state FMRI after acute stroke with hemiparesis. Neurorehabil Neural Repair 2013;27:153–163.

40.

Kuiper

Lin

Young

, et al. A parcellation-based model of the auditory network. Hear Res 2020;396:108078.

41.

Nourski

Steinschneider

McMurray

, et al. Functional organization of human auditory cortex: Investigation of response latencies through direct recordings. Neuroimage 2014;101:598–609.

42.

Morillon

Lehongre

Frackowiak

, et al. Neurophysiological origin of human brain asymmetry for speech and language. Proc Natl Acad Sci U S A 2010;107:18688–18693.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.01 MB