Brain functional connectivity asymmetry: Left hemisphere not necessarily more modular
Poster No:
1782
Submission Type:
Abstract Submission
Authors:
Lucia Jajcay1,2,3, David Tomeček1,2,3, Jiří Horáček1,4, Filip Španiel1,4, Jaroslav Hlinka1,2
Institutions:
1National Institute of Mental Health, Klecany, Czech Republic, 2Institute of Computer Science, The Czech Academy of Sciences, Prague, Czech Republic, 3Faculty of Electrical Engineering, Czech Technical University in Prague, Prague, Czech Republic, 4Third Faculty of Medicine, Charles University, Prague, Czech Republic
First Author:
Lucia Jajcay
National Institute of Mental Health|Institute of Computer Science, The Czech Academy of Sciences|Faculty of Electrical Engineering, Czech Technical University in Prague
Klecany, Czech Republic|Prague, Czech Republic|Prague, Czech Republic
National Institute of Mental Health|Institute of Computer Science, The Czech Academy of Sciences|Faculty of Electrical Engineering, Czech Technical University in Prague
Klecany, Czech Republic|Prague, Czech Republic|Prague, Czech Republic
Co-Author(s):
David Tomeček
National Institute of Mental Health|Institute of Computer Science, The Czech Academy of Sciences|Faculty of Electrical Engineering, Czech Technical University in Prague
Klecany, Czech Republic|Prague, Czech Republic|Prague, Czech Republic
National Institute of Mental Health|Institute of Computer Science, The Czech Academy of Sciences|Faculty of Electrical Engineering, Czech Technical University in Prague
Klecany, Czech Republic|Prague, Czech Republic|Prague, Czech Republic
Jiří Horáček
National Institute of Mental Health|Third Faculty of Medicine, Charles University
Klecany, Czech Republic|Prague, Czech Republic
National Institute of Mental Health|Third Faculty of Medicine, Charles University
Klecany, Czech Republic|Prague, Czech Republic
Filip Španiel
National Institute of Mental Health|Third Faculty of Medicine, Charles University
Klecany, Czech Republic|Prague, Czech Republic
National Institute of Mental Health|Third Faculty of Medicine, Charles University
Klecany, Czech Republic|Prague, Czech Republic
Jaroslav Hlinka
National Institute of Mental Health|Institute of Computer Science, The Czech Academy of Sciences
Klecany, Czech Republic|Prague, Czech Republic
National Institute of Mental Health|Institute of Computer Science, The Czech Academy of Sciences
Klecany, Czech Republic|Prague, Czech Republic
Introduction:
Recently, we examined brain functional connectivity (FC) asymmetry in terms of modularity – a statistic that quantifies the degree to which a graph (network) may be subdivided into clearly delineated groups of nodes (modules) [Newman, 2004] – and reported that the left hemisphere is more modular [Jajcay, 2022; 2023]. The finding appeared to be in line with morphological studies of the human brain and promising for studying aberrant hemispheric lateralization and its functional relevance in neuropsychiatric disorders. Here, we thus extend the original analysis (performed on healthy controls) to patients with schizophrenia. We also examine its robustness with respect to different preprocessing pipelines. Moreover, since the previous finding could potentially be due to the fact that the Automated Anatomical Labeling (AAL) atlas is not symmetrical, we further extend the analysis to additional brain parcellations.
Methods:
Using a Siemens Trio 3T MRI scanner, we measured the resting-state brain activity of 90 healthy subjects (40M – mean age 28.15 ± 6.90 years, 36 right-handed; 50F – 27.54 ± 6.82, 46) and 100 patients with schizophrenia (58M – 26.42 ± 5.05, 45; 42F – 31.75 ± 7.75, 35). Functional T2*-weighted images with blood oxygenation level-dependent (BOLD) contrast (GE-EPIs; TR/TE = 2000/30 ms, flip angle = 70°, 48×64 voxels, voxel size = 3×3×3 mm3, FOV = 192 mm, 400 volumes, 35 axial slices covering the entire cerebrum), as well as high-resolution 3D T1-weighted images (for anatomical reference), were acquired.
Using SPM12 and CONN running under MATLAB, three different data preprocessing pipelines were applied – "stringent", "moderate" (both described in Kopal [2020]), and "default" ("stringent" without linear detrending).
Individual FC matrices were computed by linear (Pearson's) correlation [Hlinka, 2011; Hartman, 2011] from – initially – the mean BOLD time series of 90 regions of interest of the AAL atlas ("AAL-90"). Using the Brain Connectivity Toolbox, the subgraphs of each hemisphere were then thresholded by preserving a proportion of the strongest weights (5–95%, in increments of 5), binarized, their modularity was computed and, finally, compared across subjects using the Wilcoxon signed-rank test.
We then repeated the analysis using additional parcellations (with "default" preprocessing). First, we created three symmetrized versions of AAL-90 by mirroring the left ("AAL-90_L") and the right ("AAL-90_R") hemispheres and by using the intersection of the mirrored masks ("AAL-90_LR"), respectively. Second, we used the gray matter areas from the MNI152 template resampled to 4284 cubes of size 6×6×6 mm³ ("cubes") as a less computationally demanding alternative to using individual voxels. Finally, since the newly created parcellations are perfectly symmetrical yet not anatomically meaningful, we also used two homotopic atlases – v2 of the volumetric AICHA which consists of 384 regions [Joliot, 2015] and the surface-based atlas by Yan et al. [2023] projected to MNI152 space, at two different resolutions ("Yan-100" and "Yan-400").
Using SPM12 and CONN running under MATLAB, three different data preprocessing pipelines were applied – "stringent", "moderate" (both described in Kopal [2020]), and "default" ("stringent" without linear detrending).
Individual FC matrices were computed by linear (Pearson's) correlation [Hlinka, 2011; Hartman, 2011] from – initially – the mean BOLD time series of 90 regions of interest of the AAL atlas ("AAL-90"). Using the Brain Connectivity Toolbox, the subgraphs of each hemisphere were then thresholded by preserving a proportion of the strongest weights (5–95%, in increments of 5), binarized, their modularity was computed and, finally, compared across subjects using the Wilcoxon signed-rank test.
We then repeated the analysis using additional parcellations (with "default" preprocessing). First, we created three symmetrized versions of AAL-90 by mirroring the left ("AAL-90_L") and the right ("AAL-90_R") hemispheres and by using the intersection of the mirrored masks ("AAL-90_LR"), respectively. Second, we used the gray matter areas from the MNI152 template resampled to 4284 cubes of size 6×6×6 mm³ ("cubes") as a less computationally demanding alternative to using individual voxels. Finally, since the newly created parcellations are perfectly symmetrical yet not anatomically meaningful, we also used two homotopic atlases – v2 of the volumetric AICHA which consists of 384 regions [Joliot, 2015] and the surface-based atlas by Yan et al. [2023] projected to MNI152 space, at two different resolutions ("Yan-100" and "Yan-400").
Results:
For a detailed overview of results across parcellations, preprocessing pipelines, and subject groups, see Table 1. Notably, the results reported previously (using AAL-90 and stringent preprocessing on healthy controls) also hold for an analogous analysis on patients with schizophrenia. However, using the more symmetrical ('artificial' or homotopic) parcellations, the results are generally not statistically significant.
Conclusions:
We have replicated our recent finding of the left hemisphere being more modular on patients with schizophrenia. However, the result appears to be sensitive to both preprocessing pipelines and brain parcellations. As a next step (in progress), we will replicate the analyses on a larger dataset (HCP) to elucidate the effect of these variables, and the nature of the asymmetry of modularity across the two hemispheres – which, for now, remains ambiguous.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia)
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural)
fMRI Connectivity and Network Modeling 1
Task-Independent and Resting-State Analysis 2
Novel Imaging Acquisition Methods:
BOLD fMRI
Keywords:
FUNCTIONAL MRI
Hemispheric Specialization
Schizophrenia
Statistical Methods
Other - Functional Connectivity, Complex Network Analysis, Graph Theory, Modularity
1|2Indicates the priority used for review
Provide references using author date format
Hlinka, J. (2011), 'Functional connectivity in resting-state fMRI: Is linear correlation sufficient?', NeuroImage, vol. 54, no. 3, pp. 2218–2225.
Jajcay, L. (2022), 'Brain functional connectivity asymmetry: Left hemisphere is more modular', Symmetry, vol. 14, no. 4, 833.
Jajcay, L. (2023, July 22–26) 'Brain functional connectivity asymmetry: Left hemisphere is more modular' [Poster no. 1511], OHBM 2023 – 29th Annual Meeting of the Organization for Human Brain Mapping, Montreal, Canada.
Joliot, M. (2015), 'AICHA: An atlas of intrinsic connectivity of homotopic areas', Journal of Neuroscience Methods, vol. 254, pp. 46–59.
Kopal, J. (2020), 'Typicality of functional connectivity robustly captures motion artifacts in rs-fMRI across datasets, atlases, and preprocessing pipelines', Human Brain Mapping, vol. 41, no. 18, pp. 5325–5340.
Newman, M.E.J. (2004), 'Finding and evaluating community structure in networks', Physical Review E, vol. 69, no. 2, 026113.
Yan, X. (2023) 'Homotopic local-global parcellation of the human cerebral cortex from resting-state functional connectivity', NeuroImage, vol. 274, 120010.