Skip to main content
. 2022 May 16;20(6):1241–1252. doi: 10.2174/1570159X19666210809101248

Table 2.

MRI scanning information, image processing method and main MRI findings for each study.

Study Scan Since ECT Images Image
Processing
Methods
Longitudinal Changes Prediction
Structural MRI Functional MRI Correlation with Symptoms
Wolf et al. [34] S1: within 5 days before ECT;
S2: 6-8 days after ECT.
T1w MRI. T1w MRI: Source-based morphometry (SBM). ECT led to an increase in structural network strength in the MTL network and left DLPFC network. No analysis. There was a significant negative correlation between PANSS total score differences and left DLPFC network strength change. No analysis.
Wang et al. [35] S1: 24 h before ECT;
S2: 24-48 h after ECT.
T1w MRI. T1w MRI: VBM analysis. ECT induced increased volumes in bilateral parahippocampal gyrus/hippocampus, right temporal gyrus, and right insula. No analysis. There was a significant positive correlation of GM volume change in the right parahippocampal gyrus/hippocampus with the reduction of positive subscore. No analysis.
Gong et al. [36] Not reported. T1w MRI, diffusion MRI. Multi-parametric MRI-based radiomic features selection and SVR model. No analysis. No analysis. No analysis. The multi-parametric MRI-based radiomic model, including four structural features from the left inferior frontal gyrus, right insula, left middle temporal gyrus and right superior temporal gyrus and six diffusion features from tracts connecting frontal or temporal gyrus, possessed the RMSE of 14.98 in the test cohort. The Pearson's correlation coefficients between predicted and actual values were 0.777.
Xi et al. [37] Not reported. T1w MRI. T1w MRI: VBM analysis and logistic regression model (LRM)/SVM. No analysis. No analysis. No analysis. The regularized multivariate LRM discriminated responders from nonresponders, with an accuracy of 90.91% in the training data and 87.59% in the validating data. The accuracy of the SVM in the training set was 90.91%, and the accuracy in the validation set was 91.78%.
Huang et al. [38] S1: 24 h before ECT;
S2: 24-48 h after ECT.
rsfMRI. rsfMRI: Voxel-wise global functional connectivity density (gFCD). No analysis. After ECT, gFCD increased in the DMN regions (medial frontal cortex and precuneus). No significant correlations. No analysis.
Li et al. [39] Not reported. rsfMRI. rsfMRI: ICA, SVM pattern classification, and FC analysis. No analysis. After ECT, the DMN (posterior cingulate cortex), the temporal lobe network (left superior temporal gyrus) and the frontal-parietal network (right angular gyrus and middle frontal gyrus) increased, meanwhile, the FC in the corticostriatal network (right anterior cingulate cortex), the language network (left middle temporal gyrus) and the DMN (right precuneus) decreased. No analysis. No analysis.
Sambataro et al. [40] S1: within 5 days before ECT;
S2: 6-8 days after ECT.
rsfMRI. rsfMRI: ICA. No analysis. ECT resulted in increased independent component loadings in the MPFC within the DMN. No significant correlations. No analysis.
Wang et al. [41] S1: 24 h before ECT;
S2: 24-48 h after ECT.
rsfMRI. rsfMRI: Seed-based FC. No analysis. ECT induced increased FC between the right thalamus and right putamen and decreased FC between the thalamus and sensory cortex. No significant correlations. No analysis.
Yang et al. [42] All patients underwent MRI scanning on the day before their first ECT session and the day after their last ECT session. rsfMRI. rsfMRI: ROI-based FC, LASSO feature selection and linear regression predictive model. No analysis. The FC between the right amygdala and left hippocampus decreased after ECT compared with the baseline. The change of FC between the amygdala and hippocampus was positively correlated with the reduction of the PANSS total score. The correlation coefficient between the predictive and real values of cross-validation was 0.7165.
Thomann et al. [43] S1: within 5 days before ECT;
S2: 6-8 days after ECT.
T1w MRI, rsfMRI. T1w MRI: VBM analysis;
rsfMRI: Seed-based FC analysis.
After ECT, increased GM volumes were found in the right amygdala, anterior part of hippocampus and insula. RUL ECT was associated with changes in FC between the right amygdala and other cortical brain regions. However, these changes were not diagnosis-specific between SZ and major depression. No significant correlations. No analysis.
Jiang et al. [44] S1: 24 h before ECT;
S2: 24-48 h after ECT.
T1w MRI, rsfMRI. T1w MRI: ROI-based GM volume analysis;
rsfMRI: Seed-based FC.
Increased GM volumes of the left insula and bilateral posterior insula (PIns) were found after ECT. After ECT, decreased FCs between the PIns and middle occipital gyrus and obitofrontal cortex were found. The PANSS reductions were correlated with the PIns volume increases and FC changes following ECT. No analysis.
Jiang et al. [45] S1: 24 h before ECT;
S2: 24-48 h after ECT.
T1w MRI, rsfMRI. T1w MRI: longitudinal segmentation of hippocampal substructures by Freesurfer;
rsfMRI: Seed-based FC.
ECT-induced volume increases in the bilateral hippocampus and certain subfields for both the remitters and nonremitters. After ECT, the remitters exhibited increased FC between the hippocampus and prefrontal cortex as well as between the hippocampus and regions in DMN. The nonremitters showed decreased FC in the hippocampus and primary sensory network. The general psychopathology reductions were associated with the left CA4 volume increase and the FC between the left caudal hippocampus and right angular gyrus. No analysis.

Abbreviations: T1w MRI: T1 weighted structural MRI, rsfMRI: resting-state functional MRI, ICA: independent component analysis, SVM: support vector machine, SVR: support vector regression, FC: functional connectivity, VBM: voxel-based morphometry, GM: gray matter, ROI: region of interest, MTL: medial temporal lobe, DLPFC: dorsal lateral prefrontal cortex, DMN: default mode network, MPFC: medial prefrontal cortex.