Table 2.
Summary of included articles analysing resting-state functional connectivity in healthy participants or patients without vascular cognitive impairment. We report imaging and clinical characteristics of patients included in each study, key steps in the acquisition and pre-processing of BOLD data, analysis of functional connectivity, and FC patterns found to be associated with CSVD. Descriptive statistics as extracted from articles are reported as range (min–max) and/or mean ± standard deviation. Missing information is indicted by empty brackets ([]). Reported are clinical characteristics of patients included in each study, details about the quantification of white matter hyperintensities, key steps in the analysis of functional connectivity, and FC patterns found to be associated with CSVD. Arrows indicate increased (↑) or reduced (↓) values, as well as positive (↗) and negative (↘) associations
| Reference | Participants | Quantification of WMH load | rs-fMRI acquisition parameters | BOLD pre-processing | FC analysis | FC patterns associated with CSVD |
|---|---|---|---|---|---|---|
| [69] |
12 depression 12 HC |
Fuzzy connected algorithm [70] |
GE Signa, 1.5 T TR 2000 ms, TE 35 ms 64 × 64 × 26, 3.75 × 3.75 × 3.8 mm3 [volumes], eyes open |
AFNI [Confound regression] [Motion scrubbing] |
PCC from ext. template SCA to define DMN Pearson correlation |
FC in DMN ↘ WMH in medial PFC |
| [71] |
47 depression 46 HC |
Not reported |
Siemens Trio, 3 T TR 2000 ms, TE 32 ms 128 × 128 × 28, 2 × 2 × 2 mm3 150 volumes, eyes open |
SPM 5 [Confound regression] [Motion scrubbing] |
PCC from ext. template SCA to define DMN Pearson correlation |
↓ Association between DMN-FC and treatment response after controlling for WMH load |
| [62] |
13 early AD 17 MCI 14 HC |
Semi-automatic using FireVoxel [72] |
[Scanner] TR 3000 ms, TE 30 ms [matrix], 3.3 × 3.3 × 3.3 mm3 140 volumes, [eyes] |
[Confound regression] [Motion scrubbing] |
Medial PFC and PCC from ext. atlas SCA to define DMN fALFF |
Both increased WMH load and reduced DMN-FC in AD and MCI compared to HC |
| [73] | 100 MCI | In-house automatic pipeline [74] |
Philips Achieva, 3 T TR 2000 ms, TE 30 ms [matrix], [resolution] 200 volumes eyes closed |
DPARSF Confound regression - 6 motion parameters - GSR: CSF, WM, global [Motion scrubbing] |
SCA from the hippocampus and PCC | No association between WMH load and FC |
| [63] |
43 MCI 24 HC |
Histogram segmentation [75] |
Training Philips, 3 T, 8-channel head coil TR 3000 ms, [TE] [matrix], 3.3 × 3.3 × 3.3 mm3 140 volumes eyes open Testing Siemens Verio, 3 T, 32-channel head coil TR 2580 ms, [TE] [matrix], 3.5 × 3.5 × 3.5 mm3 180 volumes eyes closed |
DARTEL Confound regression - 6 motion parameters - GSR: CSF, WM [Motion scrubbing] |
Whole-brain SCA Pearson correlation |
No association between WMH and FC |
| [65] |
90 MCI 140 HC |
SPM Lesion Segmentation Tool [76] |
Siemens Trio, 3 T TR 2300 ms, TE 30 ms [] × [] × 34, 3 × 3 × 4 mm3 [volumes] eyes closed |
CONN, SPM 12 Confound regression - compCor - GSR: CSF, WM Motion scrubbing - Artefact Detection Tools - Spike regression (FD > 0.5 mm) |
Preselected cognitive control networks (FPCN, SN) + DMN Pearson correlation Structural equation modelling |
Weaker negative association between executive function/memory and WMH load in patients with high global FC |
| [77] |
18 svMCI + depression 17 svMCI − depression 23 HC |
Fazekas scale |
GE MR750, 3 T TR 2000 ms, TE 35 ms 64 × 64 × 26, 4 × 4 × 4 mm3 240 volumes [eyes] |
DPABI Confound regression - Linear and quadratic trends, 24p - GSR: CSF, WM, global Motion scrubbing - Head motion > 3 mm/3° - Volume censoring (FD > 0.5 mm) |
VBM SCA from altered regions Pearson correlation |
↑ FC between right middle cingulate cortex and right parahippocampal gyrus |
| [78] |
38 P w Sjogren syndrome 40 HC |
Wahlund score |
Siemens Trio, 3 T TR 2500 ms, TE 30 ms 96 × 96 × 40, 2.3 × 2.3 × 3 mm3 204 volumes eyes closed |
Matlab, DPABI Confound regression - Linear trend - GSR: CSF, WM, global Motion scrubbing - Mean FD > 0.2 mm |
SCA from hippocampi Pearson correlation |
FC ↗ WMH left hippocampus and right inf. orbital and inf. temporal gyrus |
| Healthy participants | ||||||
| [79] | 76 healthy participants | Mixture model [80] |
GE Signa, 1.5 T TR 2000 ms, TE 40 ms [] × [] × 24, [] × [] × 5 mm3 240 volumes [eyes] |
REST Confound regression - Head motion parameters - GSR: CSF,WM, global Motion scrubbing - > 58 outlier volumes (> 1.5 mm/1.5°) |
PCC from ext. template SCA to define DMN Pearson correlation |
No association between WMH load and FC Episodic memory ↗ medial PFC–left inferior parietal cortex FC in patients with low grey matter volume |
| [81] | 127 healthy (Harvard Ageing Brain Study) |
Fazekas 0–1 vs. 2–3 |
Siemens Trio, 3 T, 12-channel head coil TR 3000 ms, TE 30 ms 72 × 72 × [], 3 × 3 × 3 mm3 124 volumes eyes open |
SPM 8 Confound regression - Realignment params + derivatives - GSR: WM, CSF, global Motion scrubbing - Mean FD > 0.15 mm - ≥ 20 outlier volumes (> 0.75 mm/1.5°) |
PCC and medial PFC from external DMN template Partial Pearson correlation Probabilistic tractography |
↓ Association between PCC-medial PFC FC and mean diffusivity in cingulum bundle |
| [82] | 186 clinically healthy (Harvard Ageing Brain Study) | Automated fuzzy-connected algorithm [70] |
Simens Trio, 3 T, 12-channel head coil TR 3000 ms, TE 30 ms 72 × 72 × 47, 3 × 3 × 3 mm3 2 × 124 volumes eyes open |
SPM 8 Confound regression - 12 motion parameters Motion scrubbing - ‘mean movement’ > 0.2 mm - > 20 outlier volumes (> 0.75 mm/1.5°) |
Template-based rotation to define DMN and FPCN Pearson correlation |
No association between WMH load and FC |
| [83] | 51 healthy participants | SPM Lesion Segmentation Tool [76] |
Phillips Ingenia, 3 T TR 2600 ms, TE 35 ms 128 × 128 × 35, 1.8 × 1.8 × 4 mm3 125 volumes, [eyes] |
REST, GIFT [Confound regression] [Motion scrubbing] |
ICA to define DMN, SN, FPN, VN |
FC in DMN ↗ WMH in the mediotemporal complex FC in SN ↗ WMH in the right S1 and sup./inf. parietal cortex |
| [84] |
1584 healthy participants (Rotterdam Study) |
Tract-specific WMH load [85] |
GE Signa, 1.5 T TR 2900 ms, TE 60 ms 64 × 64 × 31, 3.3 × 3.3 × 3.3 mm3 160 volumes, eyes open |
FSL Confound regression - Low-frequency drifts - Motion components - ICA Motion scrubbing - Max FD > 0.5 mm, abs. motion > 3 mm |
Desikan–Killiany parcellation Pearson correlation Probabilistic tractography |
FC ↘ WMH both tract-specific and global |
| [86] | 145 healthy participants | SPM Lesion Segmentation Tool [76] |
GE MR750, 3 T TR 1500 ms, TE 27 ms 64 × 64 × 29, 3.75 × 3.75 × 4 mm3 162 volumes eyes open |
FSL Confound regression - GSR: CSF, WM Motion scrubbing - FD > 0.5 mm |
ICA to define DMN, SMN, FPCN Pearson correlation |
No association between WMH load and FC |
| [87] | 69 healthy participants | Coarse-to-fine in-house developed mathematical morphology method [88] |
Phillips Achieva, 3 T TR 2050 ms, TE 25 ms 64 × 64 × 47, 3.2 × 3.2 × 3.2 mm3 210 volumes eyes open |
CONN Confound regression - 6 motion parameters - GSR: WM, CSF [Motion scrubbing] |
AAL atlas, DTI atlas Whole-brain SCA Intrinsic connectivity contrast |
FC in the left cuneus and right sup. occipital cortex ↗ WMH in the right ant. corona radiata FC in the left superior occipital cortex ↗ WMH in the right superior corona radiata |
| [67] |
400 healthy participants (Baltimore Longitudinal Study of Aging) |
Multimodal supervised classification algorithm [89] |
Philips Achieva, 3 T TR 2000 ms, TE 30 ms [matrix], 3 × 3 × 3 mm3 180 volumes [eyes] |
Confound regression - 24 motion parameters - GSR: global, WM, CSF Motion scrubbing - ‘summary motion value’ > 0.2 mm - Volume censoring (FD > 0.5 mm, < 5 min) |
Geodesic graph-based segmentation Regional homogeneity Sparse connectivity patterns |
Pattern of advanced brain ageing characterised by both increased WMH burden and reduced FC compared to resilient agers |
| [66] | 11 healthy participants | Automated regression algorithm [90] using a Hidden Markov Random Field with Expectation Maximization [91] |
Siemens Trio, 3 T TR 2000 ms, TE 27 ms 92 × 92 × 43, 2.5 × 2.5 × 3 mm3 240 volumes eyes closed |
SPM12 Confound regression - Linear/quadratic, 18 motion parameters - GSR: CSF, WM Motion scrubbing -> 3 mm max, > 3° max -> 24 spikes (FD > 1 mm) |
Brainnetome atlas (228) Graph theory to define DMN Pearson correlation |
No association between WMH load and DMN FC trajectories |
| [92] | 562 healthy participants | SPM Lesion Segmentation Tool [76] |
Phillips Achieva, 3 T TR 2000 ms, TE 20 ms 112 × 112 × 37, 2 × 2 × 3 mm3 [volumes], [eyes] |
[Confound regression] [Motion scrubbing] Mean FD as covariate in analysis |
Desikan–Killiany parcellation FC measure not specified |
No association between WMH load and FC |
| [93] |
182 participants (UK Biobank) |
BIANCA with manual correction [94] |
Siemens Skyra, 3 T TR 735 ms, TE 39 ms 88 × 88 × 64, 2.4 × 2.4 × 2.4 mm3 490 volumes, [eyes] |
FMRIB (FSL), ICA-FIX Confound regression - ICA [Motion scrubbing] |
ICA, AAL atlas Pearson correlation Degree centrality |
FC ↗ WMH in right orbitofrontal cortex |
| [95] |
250 healthy (Harvard Aging Brain Study) |
Automated fuzzy-connected algorithm [70] |
Siemens Trio, 3 T TR 3000 ms, TE 30 ms [matrix], 3 × 3 × 3 mm3 2 × 124 volumes, eyes open |
SPM 8 Template-based rotation method [Confound regression] [Motion scrubbing] |
Template-based rotation to define DMN, SMN, DMN, and FPCN Pearson correlation |
Association between WMH load and FC not investigated FC in DMN ↘ risk of progression to MCI |