Table 3.
Summary of studies assessing time-varying resting state functional connectivity modifications in different psychiatric and neurological pathologies (excluding multiple sclerosis).
| Study | RS fMRI acquisition parametersΩ | TVC analysis approach∧ | Study subjectsθ | Main findings |
|---|---|---|---|---|
| Abrol et al. (2017) | Six sites: Siemens Tim Trio 3T One site: GE Discovery MR750 3T 162 volumes TR = 2 s |
1. Group ICA decomposition in 47 relevant independent components of interest, classified into 7 functional networks 2. Sliding-window analysis, window length = 22 TRs (44 s), steps = 1 TR (2 s) 4. Clustering (five recurring states) performed using temporal ICA 3. Correlations with gray matter volumes |
FBIRN Data Repository 151 schizophrenia patients 37 females (24.5%) mean age = 37.8 years 163 healthy subjects 46 females (28.2%) mean age = 36.9 years |
- Compared to healthy subjects, patients with schizophrenia exhibited higher TVC strength between: i) sensorimotor, precuneus and parietal areas; and ii) frontal, temporal and insular cortices - In patients, TVC abnormalities correlated with lower gray matter volumes |
| Alderson et al. (2018) | Philips Intera MR 3T 140 volumes TR = 3 s |
1. Group ICA decomposition in nine relevant independent components of interest, subsequently segmented in 148 cortical regions from the Destrieux atlas (Destrieux et al., 2010) 2. Time-frequency analysis 3. Graph theory: synchrony, global metastability, eigenvector centrality, clustering coefficient, local efficiency, and participation coefficient 4. Correlation with structural abnormalities |
ADNI database 34 patients with Alzheimer's disease 18 females (52.9%) mean age = 73.79 years SD = 6.14 years 33 patients with mild cognitive impairment 13 females (39.4%) mean age = 73.61 years SD = 5.6 years 36 healthy controls 19 females (52.8%) mean age = 74.46 years SD = 5.51 years |
- In Alzheimer's disease patients, reduced synchrony was observed between right fronto-parietal regions, sensorimotor regions and DMN, together with overall reduced metastability - In patients, increased eigenvector centrality, clustering coefficient, local efficiency, and participation coefficient correlated with more severe structural damage |
| Cai J. et al. (2018) | Siemens Trio 3T Unspecified volumes TR = 2 s |
1. Segmentation of 76 brain regions from the Desikan atlas (Desikan et al., 2006) 2. Sliding-window analysis, window length = 30 TRs (60 s), steps = 2 TRs (4 s) 3. Graph theoretical analysis: global efficiency, clustering efficiency, modularity, assortativity, Fiedler value |
69 Parkinson's disease patients 30 females (43.5%) mean age = 60 years SD = 9.8 years 29 healthy controls 13 females (43.5%) mean age = 58.3 years SD = 97.5 years |
- Compared to healthy subjects, patients with Parkinson's disease showed lower network connections (Fiedler value), modularity and global efficiency - Lower network connections in patients with Parkinson's disease correlated with disease severity |
| Cetin et al. (2016) | Siemens Trio 3T 149 volumes TR = 2 s |
1. Group ICA decomposition in 39 relevant independent components of interest 2. Sliding-window analysis, window length = 31 TRs (62 s), steps = 1 TRs (2 s) k-means clustering (five recurring states) 3. Correlations with magnetoencephalography data and classification performance compared to static FC and magnetoencephalography |
47 schizophrenia patients 13 females (27.7%) mean age = 35.18 years SD = 11.83 years 45 healthy controls 7 females (15.6%) mean age = 37.28 years SD = 13.86 years |
- Classification between schizophrenia patients and healthy controls improved with TVC (accuracy = 82.79%) compared to static FC metrics (accuracy = 70.33%) - Classification performance did not improve when using a combination of TVC and magnetoencephalography metrics (accuracy = 85.35%), compared to the combination of static FC and magnetoencephalography metrics (accuracy = 87.91%) |
| Chen et al. (2018) | Philips Achieva 3T 170 volumes TR = 2 s |
1. Segmentation of left and right primary motor area, premotor cortex and supplementary motor area (spherical ROIs, radius = 5 mm) Sliding-window analysis, window length = 32TRs (64 s), steps = 1 TR (2 s) 2. Standard deviation of TVC across windows |
70 stroke patients 45 right-sided lesions 23 females (32.9%) mean age = 58.44 years SD = 11.43 years 25 left-sided lesions 8 females (11.4%) mean age = 59.88 years SD = 12.96 years 55 healthy controls 26 females (37.1%) mean age = 56.73 years SD = 10.21 years |
- Compared to healthy controls, stroke patients showed TVC reductions between sensorimotor and visual-related cortices and between the sensorimotor and the limbic system - In stroke patients with right-sided lesions, reduced TVC between the right primary motor area and the left precentral gyrus correlated with more severe disability |
| Damaraju et al. (2014) | 6 sites: Siemens Tim Trio 3T 1 site: GE Discovery MR750 3T 162 volumes TR = 2 s |
1. Group ICA decomposition in 50 relevant independent components of interest, classified into 7 different functional networks 2. Sliding-window analysis, window length = 22 TRs (44 s), steps = 1 TR (2 s) 3. k-means clustering (five recurring states) |
151 schizophrenia patients 37 females (24.5%) mean age = 37.8 years 163 healthy subjects 46 females (28.2%) mean age = 36.9 years |
- Compared to healthy controls, schizophrenia patients showed: (i) higher dwell time in states characterized by overall low inter- and intra-network TVC strength; (ii) lower dwell time in states characterized by high correlations between visual, motor and auditory networks; and (iii) increased TVC between thalami and sensory networks |
| Diez-Cirarda et al. (2018) | Philips Achieva TX 3T 214 volumes TR = 2.1 s |
1. Group ICA decomposition in 29 relevant independent components of interest, classified into seven functional networks 2. Sliding-window analysis, window length = 22 TRs (44.2 s), steps = 1 TR (2.1 s) 3. k-means clustering (two recurring states). Graph theory: global efficiency, local efficiency, clustering coefficient, betweenness centrality |
37 patients with Parkinson's disease 12 with normal cognition 6 females (50%) mean age = 65.17 years SD = 8.31 years 23 with mild cognitive impairment 10 females (44%) mean age = 69.17 years SD = 4.48 years 26 healthy controls 8 females (31%) mean age = 68.31 years SD = 7.52 years |
- Compared to healthy controls, Parkinson's disease patients with mild cognitive impairment showed lower dwell time in a state characterized by overall low strength of inter- and intra-network connections, as well as higher number of transitions between states - Parkinson's disease patients with cognitive impairment also showed: (i) reduced clustering coefficient in the right precentral gyrus vs. healthy controls; and (ii) reduced betweenness centrality of the left paracentral gyrus vs. patients without cognitive impairment |
| Du et al. (2017) | 3 sites: Siemens Trio Tim 3T 2 sites: GE Signa HDx 3T 1 site: Siemens Allegra 3T 1 site: Philips 3T 100–210 volumes TR ranging from 1.5 to 3 s |
1. Segmentation of 116 brain regions from the AAL atlas (Tzourio-Mazoyer et al., 2002) 2. Sliding-window analysis, window length = 20 TRs (ranging from 30 to 60 s) 3. GIG-ICA clustering (five recurring-states) 4. Correlations with cognitive scores |
Bipolar and schizophrenia network on intermediate phenotypes 113 schizophrenia patients 57 females (50%) mean age = 35.57 years SD = 12.29 years 132 schizoaffective disorder patients 75 females (57%) mean age = 36.23 years SD = 12.23 years 140 bipolar disorder with psychosis patients 87 females (62%) mean age = 36 years SD = 12.57 years 238 healthy controls 138 females (58%) mean age = 38.15 years SD = 12.55 years |
- Compared to healthy controls (and bipolar patients), schizophrenia and schizoaffective disorder patients showed increased TVC between frontal with angular and postcentral areas, and reduced TVC between temporal and frontal areas - Compared with all remaining study groups, schizophrenia patients also showed reduced TVC between the cerebellum and subcortical and frontal areas - Reduced TVC between cerebellar and frontal areas correlated with higher symptom severity scores |
| Du et al. (2018) | Siemens Tim Trio 3T 180 volumes TR = 2 s |
1. Segmentation of 116 brain regions from the AAL atlas (Tzourio-Mazoyer et al., 2002) 2. Sliding-window analysis, window length = 20 TRs (40 s), steps = 1 TR (2 s) 3. GIG-ICA clustering (five recurring-states) |
58 schizophrenia patients 20 females (35%) mean age = 21.8 years SD = 3.8 years 53 adults at high risk of developing schizophrenia 21 females (38%) mean age = 20.4 years SD = 4.5 years 70 healthy controls 29 females (41%) mean age = 21.9 years SD = 5.6 years |
- Compared to healthy controls, schizophrenia patients and adults with high risk of developing schizophrenia showed TVC alterations between motor, temporal, cerebellar, frontal and thalamic areas - Schizophrenia patients, compared to adults with high risk of developing schizophrenia, also showed increased TVC between the cerebellum, temporal cortex, frontal gyri and thalami - Increased TVC between temporal and cerebellar areas correlated with higher symptom severity scores |
| Engels et al. (2018) | GE Signa HDxT 3T 202 volumes TR = 2.15 s |
1. Segmentation of 264 brain regions from the Power atlas (Power et al., 2011) 2. Sliding-window analysis, window length = 28 TRs (60.2 s), steps = 5 TRs (10.75 s) 3. Standard deviation of TVC across windows |
24 Parkinson's disease patients 7 females (29.2%) mean age = 63.42 years SD = 7.93 years 27 healthy controls 11 females (40.1%) mean age = 59.37 years SD = 8.54 years |
- Compared with patients without cognitive impairment, Parkinson's disease patients with mild cognitive impairment showed higher TVC between the DMN and the rest of the brain - In patients, no correlation was found between TVC abnormalities and motor severity |
| Falahpour et al. (2016) | 17 sites TR = 2 s |
1. Manual segmentation of 10 spherical ROIs (radius = 6 and 10 mm) 2. Sliding-window analysis, window length = 15 TRs (30 s), steps = 4 TRs (8 s) 3. Standard deviation of TVC across windows |
Autism Brain Imaging Data Exchange (ABIDE) 76 autism spectrum disorders 9 females (11.8%) mean age = 16.1 years SD = 4.9 years range = 7–29.9 years 76 typically development young adults 12 females (15.8%) mean age = 15.8 years SD = 4.5 years range = 8–29.9 years |
- No between-group differences were observed in TVC |
| Fu et al. (2018) | 6 sites: Siemens Tim Trio 3T 1 site: GE Discovery MR750 3T 162 volumes TR = 2 s |
1. Group ICA decomposition in 48 relevant independent components of interest, classified into seven functional networks 2. Sliding-window analysis, window length = 20 TRs (40 s), steps = 1 TR (2 s) 3. k-means clustering of dynamic amplitude of low-frequency fluctuations (six recurring states) |
FBIRN Data Repository 151 schizophrenia patients 37 females (24.5%) mean age = 37.8 years SD = 11.4 years 163 healthy controls 46 females (28.2%) mean age = 36.9 years SD = 11 years |
- Compared to healthy controls, schizophrenia patients showed increased dynamic amplitude of low-frequency fluctuations in states characterized by strong TVC between the thalami and sensory regions - Patients also showed reduced dynamic amplitude of low-frequency fluctuations in states characterized by weak TVC between the thalami and sensory regions |
| Guo et al. (2018) | 14 sites | 1. Manual segmentation of three spherical ROIs (radius = 6 mm) 2. Flexible least squares to construct a TVC map at each timepoint 3. k-means clustering (five recurring states) Correlations with clinical scores |
Autism Brain Imaging Data Exchange (ABIDE) 209 autism spectrum disorder adolescents 0 females (0%) mean age = 16.5 years SD = 6.2 years 298 typical development adolescents 0 females (0%) mean age = 16.8 years SD = 6.2 years |
- Compared to typically developing adolescents, autism spectrum disorder adolescents showed reduced TVC among the right anterior insula, ventromedial prefrontal cortex and the posterior central cortex - Reduced TVC between the right anterior insula and the ventromedial prefrontal cortex correlated with higher symptom severity |
| He et al. (2018) | Philips Achieva 3T TR = 2 s |
1. Group ICA decomposition of the DMN, used to select the PCC for subsequent analyses 2. Sliding-window analysis, window length = 50 TRs (100 s), steps = 2 TRs (4 s) 3. Calculation of TVC map between the PCC and the rest of the brain in each window; calculation of variance of FC across windows k-means clustering analysis (six recurring states) Correlation with social behavior scales |
Autism Brain Imaging Data Exchange (ABIDE) 38 autism spectrum disorders 0 females (0%) age range = 3–7 years 41 typical development children 0 females (0%) age range = 3–6 years |
- Compared to typically developing children, Autism spectrum disorders children showed differences in TVC variance between the PCC and: (1) the whole brain; (2) the right precentral gyrus; and (3) visual areas - In autism spectrum disorder children, lower TVC variance between the PCC and the right precentral gyrus negatively correlated with social motivation |
| Jie et al. (2018) | Philips 3T scanners 140 volumes TR ranging from 2.2 to 3.1 s |
1. Segmentation of 116 brain regions from the AAL atlas (Tzourio-Mazoyer et al., 2002) 2. Sliding-window analysis, non-overlapping windows, window length ranging from 30 to 60 s 3. Metrics of temporal and spatial variability of TVC across windows 4. Six machine-learning classification algorithms |
ADNI database 43 patients with mild cognitive impairment with late onset 17 females (39.5%) mean age = 72.1 years SD = 8.2 years 56 patients with mild cognitive impairment with early onset 35 females (62.5%) mean age = 71.1 years SD = 6.8 years 50 healthy controls 29 females (58%) mean age = 75 years SD = 6.9 years |
- Patients with early mild cognitive impairment, compared to healthy controls, showed increased TVC variability - TVC abnormalities helped to identify patients with early-onset mild cognitive impairment from patients with late-onset mild cognitive impairment and healthy controls (accuracy = 74.7 and 73.6%, respectively) |
| Jones et al. (2012) | GE Signa HDx 3T 100 volumes TR = 3 s |
1. Group ICA decomposition in 54 relevant independent components of interest, used to develop 68 cubical ROIs (edge = 10 mm) 2. Sliding-window analysis, window length = 9 TRs (27 s) and 11 TRs (33 s) 3. Graph theory: variability of modularity across windows |
28 patients with Alzheimer's disease Unspecified sex and age 892 healthy controls 438 females (49%) median years = 79 years range = 75–83 years |
- Patients with Alzheimer's disease showed lower dwell time in brain states with strong contributions of the posterior areas of the DMN, and higher dwell time in states with strong contributions of the anterior areas of the DMN |
| Klugah-Brown et al. (2018) | GE Discovery MR750 3T 250 volumes TR = 2 s |
1. Group ICA decomposition in 50 relevant independent components of interest, classified into seven functional networks 2. Sliding-window analysis, window length = 22 TRs (44 s), steps = 1 TR (2 s) 3. k-means clustering (four recurring states) Individual reconstruction of TVC states using dual regression |
19 frontal lobe epilepsy patients 9 females (47.4%) median age = 24.2 years range = 13–51 years 18 healthy controls 5 females (27.8%) median age = 23.9 years range = 11–41 years |
- Compared to healthy subjects, epilepsy patients showed reduced TVC between the fronto-parietal network and cerebellar/subcortical networks - They also spent less time in the most fundamental connectivity state - A lower dwell time in this state correlated with age of seizure onset |
| Li et al. (2018) | GE Discovery 750 3T 240 volumes TR = 2 s |
1. Segmentation of cortical brain regions from the AAL atlas (Tzourio-Mazoyer et al., 2002) 2. Sliding-window analysis, window length = 50 TRs (100 s), steps = 10 TRs (20 s) 3. Standard deviation of TVC density (proportional to the number of functional connections) across windows |
43 children with benign epilepsy (centrotemporal spikes) 19 females (44.2%) mean age = 9.61 years SD = 2.04 years 28 typically developing children 13 females (46.4%) mean age = 10 years SD = 2.31 years |
- Compared to typically developing children, epilepsy children showed decreased TVC variability in the orbital inferior frontal gyrus and increased TVC variability in the precuneus - Patients with interictal epileptiform discharges, compared to patients without interictal epileptiform discharges, showed higher TVC variability in the supramarginal gyrus - Excessive TVC variability of the precuneus correlated with a younger onset age of seizure |
| Liao et al. (2018) | Unspecified GE 3T 240 volumes TR = 2 s |
1. Segmentation of 200 brain regions using the Craddock atlas (Craddock et al., 2012) 2. Sliding-window analysis, window length = 50TRs (100 s), steps = 5TRs (10 s) 3. Graph theory analyses: network strength, network efficiency, nodal efficiency, small worldness Variance of area-under-the-curve of graph metrics 4. Correlation with suicide ideation scores |
48 major depressive disorder 37 females (77.1%) mean age = 34.8 years SD = 10.3 years 30 healthy controls 18 females (60%) mean age = 35.7 years SD = 10.2 years |
- Increased network strength and efficiency in patients with suicide ideation compared to healthy subjects and major depressed patients without suicide ideation - Patients without suicide ideation showed TVC alterations within the left middle/inferior frontal gyrus, right superior parietal gyrus, right postcentral gyrus and right fusiform gyrus - TVC network strength distinguished patients with and without suicide ideation from healthy subjects |
| Liu et al. (2017) | Siemens Trio 3T 250 volumes TR = 2 s |
1. Group ICA decomposition in 21 relevant independent components of interest 2. Sliding-window analysis, window length = 55 TRs (110 s), steps = 1 TR (2 s) 3. k-means clustering (six recurring states) 4. Correlations with disease duration |
43 patients with idiopathic generalized epilepsy 15 females (34.8%) mean age = 23.12 years SD = 4.8 years 48 healthy controls 19 females (39.5%) mean age = 23.02 years SD = 1.49 years |
- Patients with idiopathic generalized epilepsy showed reduced dwell time in a state characterized by strong correlations between visual and remaining sense-related networks, as well as increased dwell time in a state characterized by strong correlations between cognitive and sense-related networks - In patients with idiopathic generalized epilepsy, reduced dwell time in the first above-mentioned state was correlated with a higher seizure frequency |
| Liu et al. (2018) | Siemens Trio 3T Unspecified volumes TR = 2 s |
1. Segmentation of bilateral putamen and 56 brain regions from the Desikan atlas (Desikan et al., 2006) 2. Sliding-window analysis, window length = 30 TRs (60 s), steps = 2 TRs (4 s) 3. Standard deviation of TVC strength 4. Correlations with clinical scores |
30 patients with Parkinson's disease 11 females (36.7%) mean age = 57.8 years SD = 9.9 years 28 healthy controls 14 females (50%) mean age = 58.4 SD = 7.6 years |
- Compared to healthy controls, Parkinson's disease patients showed reduced TVC between the posterior subunit in the left putamen with the left superior frontal gyrus, right putamen and the right precentral gyrus, as well as between the right posterior putamen and bilateral pallidum nuclei - TVC abnormalities correlated with more severe disability |
| Mennigen et al. (2018) | Siemens Trio 3T 170 volumes TR = 2 s |
1. Group ICA decomposition in 47 relevant independent components of interest, classified into eight functional networks 2. Sliding-window analysis, window length = 22 TRs (44 s), steps = 1 TR (2 s) 3. k-means clustering (five recurring states). Fuzzy meta-state analysis |
53 patients with clinical high-risk for psychosis 21 females (39.6%) mean age = 20.4 years SD = 4.5 years 58 schizophrenia patients 20 females (34.5%) mean age = 21.8 years SD = 3.8 years 70 healthy controls 41 females (58.6%) mean age = 21.9 years SD = 5.6 years |
- Compared to healthy subjects, schizophrenia patients showed significantly lower global meta-state dynamism - Compared to healthy controls, patients with high-risk for psychosis showed significantly lower meta-state dynamism |
| Qiu et al. (2018) | GE Excite 3T 195 volumes TR = 2 s |
1. Segmentation of three amygdalar subregions in each hemisphere, following the JuBrain Cytoarchitectonic Atlas (Zilles and Amunts, 2010) 2. Sliding-window analysis, window length = 100 TRs (200 s), step = 1 TR (2 s) 3. Voxel-wise maps of variance of amygdalar TVC across windows |
30 patients with major depression disorder 20 females (66.7%) mean age = 36.1 years SD = 12.3 years range = 18–60 years 62 healthy controls 33 females (53.2%) mean age = 35.1 years - SD = 15.9 years - range = 16–81 years |
- Compared to healthy controls, patients with major depression disorder exhibited decreased positive TVC correlations between the amygdala and left centromedial and superficial subregions, primarily in the brainstem, decreased positive fronto-thalamic TVC, and decreased negative TVC of the left centromedial subregion with the right superior frontal gyrus - In patients, mean positive TVC strength between the left centromedial region and brainstem was positively correlated with the age of onset of major depression disorder |
| Quevenco et al. (2017) | Philips Achieva 7T 200 volumes TR = 2 s |
1. Segmentation of 90 brain cortical regions from the AAL atlas (Tzourio-Mazoyer et al., 2002) 2. Sliding-window analysis, window length = 30 TRs (60 s), steps = 1 TR (2 s) 3. Principal components analysis: eigen-connectivity patterns (states) |
37 healthy controls divided according to presence/absence of memory decline 13 females (35.1%) mean age = 73 years SD = 6.6 years |
- Subjects with memory decline showed reduced TVC between anterior and posterior brain areas - Increased global connectivity, reduced TVC between anterior and posterior brain areas, increased TVC between interhemispheric fronto-temporal areas and reduced TVC between parietal and temporal areas correlated with memory decline and apoprotein E-ε4 carrier status |
| Rashid et al. (2014) | Siemens Allegra 3T Eyes open 202 volumes TR = 1.5 s |
1. Group ICA decomposition in 49 relevant independent components of interest, classified into 7 functional networks 2. Sliding-window analysis, window length = 22 TRs (33 s), steps = 1 TR (1.5 s) 3. k-means clustering (five recurring states) |
60 schizophrenia patients 13 females (21.7%) mean age = 35.85 years SD = 12.01 years 38 bipolar disorder patients 20 females (52.6%) mean age = 38.96 years SD = 10.9 years 61 healthy controls 28 females (45.9%) mean age = 35.4 years SD = 11.57 years |
- Compared to controls, schizophrenia patients showed increased TVC between: (i) temporal regions; (ii) frontal regions; (iii) subcortical regions; iv) temporal and parietal regions, and reduced TVC between: (i) frontal and parietal regions and (ii) frontal and occipital areas - Compared to bipolar patients, schizophrenia patients showed increased TVC between: (i) frontal and parietal areas; (ii) sensorimotor areas; (iii) sensorimotor and parietal areas - Compared to healthy controls, bipolar disorder patients showed increased TVC between temporal and parietal areas, as well as reduced TVC within parietal regions |
| Rashid et al. (2016) | Siemens Allegra 3T Eyes open 202 volumes TR = 1.5 s |
1. Group ICA decomposition in 49 relevant independent components of interest, classified into seven functional networks 2. Sliding-window analysis, window length = 22 TRs (33 s), steps = 1 TR (1.5 s) 3. k-means clustering (five recurring-states) 4. Machine learning classification of the study subgroups |
60 schizophrenia patients 13 females (21.7%) mean age = 35.85 years SD = 12.01 years 38 bipolar disorder patients 20 females (52.6%) mean age = 38.96 years SD = 10.9 years 61 healthy controls 28 females (45.9%) mean age = 35.4 years SD = 11.57 years |
- TVC improved classification between patients with schizophrenia, patients with bipolar disorder and healthy controls: TVC overall classification accuracy (84.28%) was significantly higher than overall classification accuracy of static FC metrics (59.12%) |
| Rashid et al. (2018a) | GE Discovery 3T 160 volumes TR = 2 s |
1. Group ICA decomposition in 38 relevant independent components of interest 2. Sliding-window analysis, window length = 22 TRs (44 s), steps = 1 TR (2 s) 3. k-means clustering (four recurring states) 4. Sex and age association with recurring-states |
Generation R study 774 children 22 children diagnosed with autism spectrum disorders 15 children with autistic traits age range = 4.89–8.90 years 774 typical development children |
- In typically developing children, TVC globally increased with age in fronto-temporal, fronto-parietal and temporo-parietal networks - Compared to typically developing children, autism spectrum disorder children showed: (i) increased TVC between the right insula and left superior frontal gyrus, right supramarginal gyrus and left precuneus; and (ii) reduced TVC between the right insula and the right supramarginal gyrus, the left supplementary motor area and right supramarginal gyrus - Autism spectrum disorder patients with high level of autistic traits showed longer dwell times in a globally disconnected state |
| Rashid et al. (2018b) | Siemens Trio 3T GE Discovery MR750 3T 162 volumes TR = 2 s |
1. Group ICA decomposition in 7 relevant independent components of interest 2. Sliding-window analysis, window length = 22 TRs (44 s), steps = 1 TR (2 s) 3. k-means clustering (five recurring states) Correlation with peak weights of single nucleotide polymorphism mostly located in chromosome 6 |
FBIRN Data Repository 61 schizophrenia patients 9 females (14.8%) mean age = 38.4 years 87 healthy controls 26 females (29.9%) mean age = 36.8 years |
- Schizophrenia patients showed a lower occupancy rate of a state characterized by high TVC in temporal, parietal, limbic and occipital regions (state 1), as well as a higher occupancy rate of a state characterized by increased fronto-limbic and intra-occipital TVC (state 5) vs. healthy subjects - Schizophrenia patients with increased gene polymorphism had stronger disrupted TVC in states 1 and 5 |
| Ridley et al. (2017) | Siemens Avanto 1.5T 200 volumes TR = 3 s |
1. Segmentation of spherical ROIs (radius = 5 mm), defined by their contact to implanted electrodes 2. Sliding-window analysis, window length = 30 TRs (90 s), steps = 0.66 TR (2 s) 3. Correlation with EEG data |
9 patients with drug-resistant epilepsy 3 females (33.3%) mean age = 30.4 years SD = 4.5 years range = 24–38 years No control group |
- In cortices not involved by epilepsy, TVC was correlated with EEG registration of all frequency bands - In epileptic cortices, TVC correlated with EEG in alpha band |
| Sakoglu et al. (2010) | Siemens Allegra 3T Active fMRI (auditory oddball task): Two consecutive runs 249 volumes TR = 1.5 s |
1. Group ICA decomposition in 10 relevant independent components of interest 2. Sliding-window analysis, window length = 64 TRs (96 s), steps = 2 TRs (3 s) 3. Time-frequency analysis 4. Standard deviation of TVC across windows, between-group comparison of TVC in each window |
28 schizophrenia patients 5 females (17.9%) mean age = 36.4 years SD = 12.43 years 28 healthy controls 9 females (32.1%) mean age = 28.8 years SD = 10.7 years |
- Compared to controls, schizophrenia patients exhibited reduced TVC task-modulation between the medial temporal network and the right lateral fronto-parietal/frontal networks. They also showed increased TVC task-modulation between the motor and frontal networks, and between the posterior DMN and orbitofrontal/parietal networks |
| Sun et al. (2018) |
Laboratory dataset
Philips Achieva 3T Eyes open 240 volumes TR = 2 s COBRE dataset Siemens Trio 3T 150 volumes TR = 2 s |
1. Segmentation of 90 brain regions from the AAL atlas (Tzourio-Mazoyer et al., 2002) 2. Sliding-window analysis, window length = 50 TRs (100 s), steps = 3 TRs (6 s) 3. Graph theory analysis: temporal global/local efficiency, richness, sparsity range of temporal networks |
Laboratory dataset 18 schizophrenia patients 8 females (44.4%) mean age = 38.8 years SD = 9.9 years range = 24–56 years 19 healthy controls 9 females (47.4%) mean age = 37.7 years SD = 9.0 years range = 28–59 years COBRE dataset 53 schizophrenia patients 12 females (22.6%) mean age = 38.3 years SD = 13.9 years range = 18–65 years 57 healthy controls 20 females (35.1%) mean age = 35.4 years SD = 11.9 years range = 18–62 years |
- Compared to healthy controls, schizophrenia patients showed higher temporal regional efficiency with left frontal, right medial parietal and bilateral subcortical areas - Abnormalities of temporal network efficiency correlated with a higher presence of schizophrenia positive and negative symptoms |
| Vergara et al. (2018) | Siemens Trio 3T 145 volumes TR = 2 s |
1. Group ICA decomposition in 48 relevant independent components of interest, classified in nine functional networks 2. Sliding-window analysis, window length = 15 TRs (30 s) 3. k-means clustering (four recurring states) 4. Machine learning for group classification |
48 patients with mild traumatic brain injury 25 females (52.1%) mean age = 27.79 years SD = 9.18 years 48 healthy controls 25 females (52.1%) mean age = 27.40 years SD = 8.96 years |
- Compared to healthy controls, mild traumatic brain injury patients showed stronger TVC between the cerebellum and sensorimotor areas, as well as a trend toward increased connectivity between the cerebellum and almost all cortical areas - Results were similar to those obtained with the study of static FC (Vergara et al., 2017) |
| Wang et al. (2018) | Siemens TIM Trio 3T 300 volumes TR = 2 s |
1. Voxel-by-voxel calculation of connection strength index (CSI) and connection count index (CCI) within a whole gray matter from the MNI template (Evans et al., 1992) 2. Sliding-window analysis, window length = 60 TRs (120 s), steps = 1 TR (2 s) 3. Mean of CSI and CCI across windows |
18 patients with juvenile myoclonic epilepsy 15 females (83.3%) mean age = 30.11 years SD = 7.73 years range = 20-48 years 25 young adults 10 females (40%) mean age = 33.2 years SD = 13.5 years |
- Patients with juvenile myoclonic epilepsy showed increased TVC in the left dorsolateral prefrontal cortex, dorsal striatum, precentral and middle temporal gyri |
| Yaesoubi et al. (2017a) | Siemens Tim Trio 3T GE Discovery MR750 3T 162 volumes TR = 2 s |
1. Group ICA decomposition in 50 relevant independent components of interest, using data from a subgroup of 120 healthy subjects 2. Time-frequency analysis 3. k-means clustering (five recurring states) |
FBIRN Data Repository 163 healthy subjects 46 females (28.2%) mean age = 36.9 years 151 schizophrenia patients 37 females (24.5%) mean age = 37.8 years |
- Using temporal and frequency information, it was possible to estimate TVC states present both in healthy controls and schizophrenia patients (characterized by very high or very low frequency profiles), and states present just in one group - Compared to controls, schizophrenia patients showed more connectivity patterns characterized by anti-correlations between the sensorimotor and visual/auditory/subcortical networks, as well as more lagged correlation between the DMN and sensory networks |
| Yu et al. (2015) | Siemens Trio 3T Eyes open 150 volumes TR = 2 s |
1. Group ICA decomposition in 48 relevant independent components of interest, classified into six functional networks 2. Sliding-window analysis, window length = 20 TRs (40 s), step = 1 TR (2 s) 3. Graph theory: connectivity strength, clustering coefficient, global efficiency; variance of graph metrics over time. 4. Assessment of reoccurring connectivity states based on graph metrics (four recurring states) |
82 schizophrenia patients 17 females (20.7%) mean age = 38 years SD = 14 years 82 healthy controls 19 females (23.2%) mean age = 37.7 years SD = 10.8 years |
- Compared to controls, schizophrenia patients showed lower connectivity strength, clustering coefficient and global efficiency, as well as higher occupancy rate of a state characterized by disconnection between the sensorimotor, the cognitive control, and the DMN |
| Yue et al. (2018) | Siemens Trio 3T 240 volumes TR = 2 s |
1. Segmentation of bilateral amygdalae, using stereotaxic and probabilistic maps of cytoarchitectonic boundaries 2. Sliding-window analysis, window length = 18 TRs (36 s) 3.Standard deviation of voxel-wise amygdalar TVC across windows |
33 schizophrenia patients 22 females (66.7%) mean age = 30.6 years SD = 8.13 years 34 healthy controls 20 females (58.8%) mean age = 28.12 years SD = 6.5 years |
- Compared to controls, schizophrenia patients showed increased TVC between the left amygdala and orbitofrontal regions - In schizophrenia patients, variability of TVC correlated with worse information processing and attention performance, as well as with more severe disease severity |
| Zhang W. et al. (2018) | Siemens Trio 3T 1,000 volumes TR = 0.427 s |
1. Segmentation of Brodmann areas 44, 45 (frontal), 22, 40 (auditory) (Zilles and Amunts, 2010) 2. Sliding-window analysis, window length 100 TRs (42.7 s), steps = 2 TRs (0.85 s) 3. k-means clustering (5 recurring states) 4. Variance of TVC strength between ROIs across windows. 5. Correlation with clinical scales |
35 schizophrenia patients 14 females (40%) mean age = 32.61 years SD = 11.58 years 22 healthy controls 13 females (60%) mean age = 34.91 years SD = 13.34 years |
- Schizophrenia patients with auditory hallucinations showed decreased TVC between the left frontal speech and left temporal auditory areas vs. healthy controls |
| Zhi et al. (2018) |
Multicenter setting
Philips Achieva 3T Siemens Verio 3T Siemens Prisma 3T 240 volumes TR = 2 s |
1. Group ICA decomposition in 49 relevant independent components of interest, classified into eight functional networks 2. Sliding-window analysis, window length = 22 TRs (44 s), steps = 1 TR (2 s) 3. k-means clustering (five recurring states) 4. Graph theory: global and node properties in each connectivity state 5. Correlations with depression severity and cognitive score |
182 major depressive disorder patients 119 females (65.4%) mean age = 32.0 years SD = 10.3 years 218 healthy controls 142 females (65.2%) mean age = 29.5 years SD = 8.3 years |
- Compared to controls, major depressive disorder patients showed: (i) higher TVC strength between the superior frontal and middle frontal gyrus; (ii) decreased TVC between the lingual gyrus and middle occipital gyrus; and (iii) decreased TVC between the superior parietal lobe and middle frontal gyrus - Correlation between TVC abnormalities and: (i) more severe depressive symptoms, impaired attention and worse executive functions; (ii) lower attention; and (iii) worse performances at working memory and executive functions |
Ω
All RS scans were acquired in the eyes-closed condition, except where indicated.
∧
TVC analysis approach summarizes: (1) ROIs used; (2) assessment of time-varying correlations between brain regions; (3) features extracted for assessing TVC.
θ
For each study group of healthy subjects, sex is represented as number of females (%), mean age and standard deviation (SD).
RS, resting state; fMRI, functional magnetic resonance imaging; TVC, time-varying functional connectivity; TR, repetition time; ICA, independent component analysis; FBIRN, function biomedical informatics research network data; ADNI, Alzheimer's disease neuroimaging initiative; SD, standard deviation; DMN, default-mode network; FC, functional connectivity; ROI, region of interest; GIG-ICA, group-information-guided ICA; AAL, automated anatomical labeling; COBRE, center for biomedical research excellence; EEG, electroencephalographic registration; PCC, posterior cingulate cortex; CSI, connection strength index; CCI, connection count index; MNI, Montreal Neurological Institute.