Graph theoretical analysis of functional brain networks: test-retest evaluation on short- and long-term resting-state functional MRI data.

Wang JH, Zuo XN, Gohel S, Milham MP, Biswal BB, He Y. · 2011

Read Graph theoretical analysis of functional brain networks: test-retest evaluation on short- and long-term resting-state functional MRI data. by Wang JH, Zuo XN, Gohel S, Milham MP, Biswal BB, He Y. (2011) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc3139595.txt.

Introduction

Jin-Hui Wang and colleagues tested whether brain network maps give the same answer twice in a 2011 paper in PLoS ONE. Researchers increasingly chart the resting brain as a network of linked regions and summarise it with graph measures. The team rescanned volunteers within an hour and again after months, then checked how steady twelve global and six local measures stayed.

Global measures proved shaky overall, shifting with time gap, sign handling, and how links were weighted. Local measures varied widely by region and type, though connection count per region was the steadiest of all. Hub regions in higher brain areas showed fair steadiness. Computer simulations blamed sensitivity to noise in the link data. The paper guides analysts toward the schemes and measures that can be trusted.

Catalog record

CREATORWang JH, Zuo XN, Gohel S, Milham MP, Biswal BB, He Y.
YEAR2011
COLLECTIONPAPERS
FORMATTXT
FILE SIZE90 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC3139595

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Graph Theoretical Analysis of Functional Brain Networks: Test-Retest Evaluation on Short- and Long-Term Resting-State Functional MRI Data By Jin-Hui Wang, Xi-Nian Zuo, Suril Gohel, Michael P Milham, Bharat B Biswal, Yong He, Marcus Kaiser Published in: PLoS ONE DOI: 10.1371/journal.pone.0021976 ABSTRACT Abstract Graph-based computational network analysis has proven a powerful tool to quantitatively characterize functional architectures of the brain. However, the test-retest (TRT) reliability of graph metrics of functional networks has not been systematically examined. Here, we investigated TRT reliability of topological metrics of functional brain networks derived from resting-state functional magnetic resonance imaging data. Specifically, we evaluated both short-term (<1 hour apart) and long-term (>5 months apart) TRT reliability for 12 global and 6 local nodal network metrics. We found that reliability of global network metrics was overall low, threshold-sensitive and dependent on several factors of scanning time interval (TI, long-term>short-term), network membership (NM, networks excluding negative correlations>networks including negative correlations) and network type (NT, binarized networks>weighted networks). The dependence was modulated by another factor of node definition (ND) strategy. The local nodal reliability exhibited large variability across nodal metrics and a spatially heterogeneous distribution. Nodal degree was the most reliable metric and varied the least across the factors above. Hub regions in association and limbic/paralimbic cortices showed moderate TRT reliability. Importantly, nodal reliability was robust to above-mentioned four factors. Simulation analysis revealed that global network metrics were extremely sensitive (but varying degrees) to noise in functional connectivity and weighted networks generated numerically more reliable results in compared with binarized networks. For nodal network metrics, they showed high resistance to noise in functional connectivity and no NT related differences were found in the resistance. These findings provide important implications on how to choose reliable analytical schemes and network metrics of interest. INTRODUCTION The human brain is a highly complex system represented as a structurally interconnected network by a dense of cortico-cortical axonal pathways (i.e., structural connectome, [1]) and a functionally synchronized network by external or intrinsic coherent neural activity (i.e., functional connectome, [2]). Mapping the brain connectome and highlighting the underlying organizational principles are fundamental for our understanding of the brain architecture. Recent studies have manifested that human brain connectome networks can be constructed using neuroimaging (e.g., functional MRI (fMRI) and diffusion tensor imaging (DTI)) or electrophysiological (e.g., electroencephalography (EEG) and magnetoencephalography (MEG)) data and further investigated by graph theoretical approaches. These brain networks have consistently demonstrated many non-trivial topological properties, such as small-worldness, modularity and highly connected hubs (for reviews, see [3], [4], [5], [6], [7]), and exhibited distinct alterations associated with different neurocognitive disorders (for reviews, see [8], [9]). While graph theoretical approaches provide valuable insights into normal brain architecture and pathological mechanism for brain disorders, the test-retest reliability has not been systematically investigated. Reliable measures are fundamental to infer trustworthy conclusions and to serve as potential clinical biomarkers. In response to the demand, several groups examined the TRT reliability/reproducibility of graph network metrics. In anatomical world, Vaessen et al. [10] assessed the reproducibility of anatomical brain networks derived from DTI data and reported high inter-scan reproducibility of network metrics across sampling schemes (e.g., number of gradient directions and gradient amplitude). Bassett et al. [11] demonstrated high reproducibility and low variability of graph metrics for both DTI and diffusion spectrum imaging data derived networks. As for functional imaging arena, Deuker et al. [12] investigated the TRT reliability of functional brain networks using MEG data and reported high reliability during a working memory task but relatively low under resting condition for network metrics. More recently, Telesford et al. [13] constructed functional brain networks using baseline fMRI data during an executive task and demonstrated excellent reproducibility for both small-world properties and network efficiency metrics. Despite these progresses, however, the TRT reliability of network metrics derived from resting-state fMRI (R-fMRI) dataset has not been well documented so far. R–fMRI is a promising tool to non-invasively map intrinsic functional connectivity patterns of the human brain in vivo [2], [14], [15], [16] and has been extensively used to investigate inherent brain network topological organization (for a review, see [17]). Of note, several previous R-fMRI studies suggest that the strength of interregional functional connectivity is dynamic in time (from seconds to minutes) and frequency domains [18] and can be modulated by the levels of current conscious awareness [19], [20], [21], [22], emotional state [23] and cognitive demand prior to resting-state scanning [24], [25], [26]. Using R-fMRI, Shehzad et al. [27] have demonstrated modest to good TRT reliability for some specific functional connections. However, these states or experiments related alterations in functional connectivity may further interact with the global network topology [28]. To our best knowledge, there are no studies to systematically examine the TRT reliability of network topological metrics derived from R-fMRI data. Accordingly, systematic and direct work is clearly warranted. In the current study, we implemented a comprehensive estimation of TRT reliability for both global network properties and regional nodal characteristics of intrinsic functional brain networks constructed using a public TRT R-fMRI dataset (http://www.nitrc.org/projects/trt). This dataset allows us to examine both short-term (<1 hour apart) and long-term (>5 months apart) network reliability. Moreover, given numerous discrepancies in the analytical strategies of existing brain network studies (e.g., how to define network nodes or how to deal with negative correlations/connections), we further evaluated the effects of three factors on network reliability. They are: (1) network node definition (ND, i.e, structural regions of interest (ROIs) based node definition or functional ROIs based node definition); (2) network membership (NM, i.e., inclusion or exclusion negative correlations); and (3) network type (NT, i.e., binarized or weighted networks). Table 1 lists those acronyms specific to the current study. [Table: Brief descriptions of several specific acronyms used in the present study.] METHODS Subjects We used a TRT R-fMRI dataset of 25 participants (mean age 30.7 ± 8.8, 9 males) that is publicly available at NITRC (http://www.nitrc.org/projects/trt). The dataset has been used to examine TRT reliability of seed-based resting-state functional connectivity (RSFC) [27], independent component analysis and dual regression [29], amplitude of low-frequency fluctuations [30] and functional homotopy [31]. Data acquisition Three resting-state scans were obtained for each participant using a Siemens Allegra 3.0-Tesla scanner. Each scan consisted of 197 contiguous EPI functional volumes (time repetition (TR) = 2000 ms; time echo (TE) = 25 ms; flip angle (FA) = 90°, number of slices = 39, matrix = 64×64; field of view (FOV) = 192 mm; acquisition voxel size = 3×3×3 mm3). Scans 2 and 3 were conducted in a single-scan session, 45 minutes apart, and were 5–16 months (mean 11±4) after scan 1. All individuals were asked to relax and remain still with eyes open during the scan. Additionally, a high-resolution T1-weighted magnetization prepared gradient echo sequence was also obtained (MPRAGE, TR = 2500 ms; TE = 4.35 ms; inversion time = 900 ms; FA = 8°; number of slices = 176; FOV = 256 mm). Data preprocessing Data preprocessing was performed using SPM5 package (http://www.fil.ion.ucl.ac.uk/spm). First, all images were corrected for intra-volume acquisition time offsets between slices using the Sinc interpolation and inter-volume geometrical displacement due to head movement using six-parameter (rigid body) transformation. Then all functional images were normalized into the Montreal Neurological Institute space using an optimum

Read the complete text