Reliability Of Functional Connectivity Research Articles

The intra-session reliability of functional connectivity during naturalistic viewing conditions.

Functional connectivity analysis is a common approach to the characterization of brain function. While studies of functional connectivity have predominantly focused on resting-state fMRI, naturalistic paradigms, such as movie watching, are increasingly being used. This ecologically valid, yet relatively unconstrained acquisition state has been shown to improve subject compliance and, potentially, enhance individual differences. However, unlike the reliability of resting-state functional connectivity, the reliability of functional connectivity during naturalistic viewing has not yet been fully established. The current study investigates the intra-session reliability of functional connectivity during naturalistic viewing sessions to extend its understanding. Using fMRI data of 24 subjects measured at rest as well as during six naturalistic viewing conditions, we quantified the split-half reliability of each condition, as well as cross-condition reliabilities. We find that intra-session reliability is relatively high for all conditions. While cross-condition reliabilities are higher for pairings of two naturalistic viewing conditions, split-half reliability is highest for the resting state. Potential sources of variability across the conditions, as well as the strengths and limitations of using intra-session reliability as a measure in naturalistic viewing, are discussed.

Reliability of functional connectivity in resting-state functional MRI

Functional MRI is a noninvasive method in brain imaging. Localization, classification, prediction and connectivity are the most common issues. Functional connectivity is a branch of fMRI that focuses on connectivity between voxels and ROIs. There are several methods for investigating functional connectivity such as correlation analysis. In any field, it is very important that results of any research have reliability according to the experiment. Any methods and measurement instruments need to be reliable. Without reliability, results are meaningless and our research is not trustworthy. Brain imaging can be used as a valuable tool for pre-surgical planning, so the results should be highly reproducible. Test-retest reliability can be explored using the intra-class correlation coefficient (ICC). I2C2 is an extent of ICC to verify the reliability in high-dimensional data as imaging studies. 13 subjects of test-retest resting-state fMRI are used to investigate reliability. I2C2 of four ROIs are also computed (Caudate, Cingulate, Cuneus and Precentral regions). Functional connectivity is found to have moderate reliability ranging 0.6244 to 0.6941. 95% confidence interval of I2C2 is calculated by nonparametric bootstrap in which CI of Caudate region I2C2 has the shortest length.

EP 29. Resting-state test-retest reliability over different preprocessing steps

Functional magnetic resonance imaging (fMRI) plays a vital role in understanding normal and clinical brain function and relies on detecting changes in blood oxygenation (i.e., BOLD). Synchronized signal fluctuations can be observed even when the subject is at rest, i.e., without performing any task. Therefore, analyzing resting-state data has become one way of studying ongoing brain activity and interrelation of brain regions. Apparent brain activity can be influenced by unwanted signals including system noise, thermal noise and noise induced by non-neuronal physiological processes. The latter induced noise is mostly the unwanted noise that affects the BOLD signal. Globally influencing nuisance regressors are derived from either whole brain or specific tissues types and removed from the main signal ( Behzadi et al., 2007 , Murphy et al., 2009 ). Removal of various global nuisance regressors alters the variance of the residual signal. Functional connectivity (FC) in general measures, how well different regions in the brain relate to each other. It estimates the common variance of the signal fluctuations within these regions by linear correlation. In this study, we tested the reliability of FC after removing confounding noise regressors. Here we focused on the effects of various commonly used confound removals in the resting state studies, such as PCA de-noising, global mean signal regression including white matter (WM) and CSF mean signal regression, tissue signal regression (Grey matter (GM), WM and CSF). Additionally, we examined GM specific time series extraction from seed regions. We conducted a seed based FC analysis on 42 subjects scanned twice with an interval of 100–250 day and tested the reliability between scans. In order to compute the seed based functional connectivity, a priori defined networks were analyzed (extended socioaffective default mode network and the working-memory network). Both these networks show robust within network resting state connectivity, as well as anti-correlation between each other. The reliability of functional connectivity is measured using two different measures i.e., by computing the spearman correlations and the absolute differences of the functional connectivity scores. Our results ( Fig. 1 ) showed that GM masking of the seed regions based on the group-average GM probabilities is advisable. Also, PCA de-noising reduces the reliability of connectivity estimates. Finally, with respect to global signal regression, we observed that refraining from this approach enhances test-retest reliability but comes at the expense of potentially poorer biological validity, indicated by missing anti-correlations between what has been previously described as antagonistic networks. Here removal of global WM and CSF signals seems to provide a good compromise, as this approach yielded reliable and meaningful estimates of within and between-network connections. Importantly, reliability showed correlation with the retained variance, presumably including structured noise. Consequently, noise removal from fMRI data requires a compromise between maximizing the test-retest reliability and removing variance that may be attributable to non-neuronal sources.