Linear Mixed-Effects Modeling Approach to FMRI Group Analysis.
Scientific and Statistical Computing Core, NIMH/NIH/HHS, USA. Electronic address: .NeuroImage (Impact Factor: 6.36). 01/2013; 73. DOI: 10.1016/j.neuroimage.2013.01.047
Conventional group analysis is usually performed with Student-type t-test, regression, or standard AN(C)OVA in which the variance-covariance matrix is presumed to have a simple structure. Some correction approaches are adopted when assumptions about the covariance structure is violated. However, as experiments are designed with different degrees of sophistication, these traditional methods can become cumbersome, or even be unable to handle the situation at hand. For example, most current FMRI software packages have difficulty analyzing the following scenarios at group level: (1) taking within-subject variability into account when there are effect estimates from multiple runs or sessions; (2) continuous explanatory variables (covariates) modeling in the presence of a within-subject (repeated measures) factor, multiple subject-grouping (between-subjects) factors, or the mixture of both; (3) subject-specific adjustments in covariate modeling; (4) group analysis with estimation of hemodynamic response (HDR) function by multiple basis functions; (5) various cases of missing data in longitudinal studies; and (6) group studies involving family members or twins.
[Show abstract] [Hide abstract]
- "Linear mixed-effects (LME) models with time as a fixed factor and a random intercept to account for missing data were used to investigate changes in time as a function of recovery. Voxel-wise LMEs were performed using the AFNI program 3dLME (Chen et al. 2013). Two-tailed independent samples t-tests were used to characterize cross-sectional differences against HA. "
ABSTRACT: Growing evidence suggests that sports-related concussions (SRC) may lead to acute changes in intrinsic functional connectivity, although most studies to date have been cross-sectional in nature with relatively modest sample sizes. We longitudinally assessed changes in local and global resting state functional connectivity using metrics that do not require a priori seed or network selection (regional homogeneity; ReHo and global brain connectivity; GBC, respectively). A large sample of collegiate athletes (N = 43) was assessed approximately one day (1.74 days post-injury, N = 34), one week (8.44 days, N = 34), and one month post-concussion (32.47 days, N = 30). Healthy contact sport-athletes served as controls (N = 51). Concussed athletes showed improvement in mood symptoms at each time point (p's < 0.05), but had significantly higher mood scores than healthy athletes at every time point (p's < 0.05). In contrast, self-reported symptoms and cognitive deficits improved over time following concussion (p's < 0.001), returning to healthy levels by one week post-concussion. ReHo in sensorimotor, visual, and temporal cortices increased over time post-concussion, and was greatest at one month post-injury. Conversely, ReHo in the frontal cortex decreased over time following SRC, with the greatest decrease evident at one month post-concussion. Differences in ReHo relative to healthy athletes were primarily observed at one month post-concussion rather than the more acute time points. Contrary to our hypothesis, no significant cross-sectional or longitudinal differences in GBC were observed. These results are suggestive of a delayed onset of local connectivity changes following SRC.
[Show abstract] [Hide abstract]
- "Linear mixed-effects (LME) models with time as a fixed factor were used to investigate changes in time as a function of recovery. The AFNI program 3dLME was used to perform voxel-wise analyses on FA data [Chen et al., 2013]. For ROI analyses, three LME models (corpus callosum, left hemisphere, right hemisphere) were performed with time, ROI, and the interaction between time and ROI as fixed effects. "
ABSTRACT: There is great interest in developing physiological-based biomarkers such as diffusion tensor imaging to aid in the management of concussion, which is currently entirely dependent on clinical judgment. However, the time course for recovery of white matter abnormalities following sports-related concussion (SRC) is unknown. We collected diffusion tensor imaging and behavioral data in forty concussed collegiate athletes on average 1.64 days (T1; n = 33), 8.33 days (T2; n = 30), and 32.15 days post-concussion (T3; n = 26), with healthy collegiate contact-sport athletes (HA) serving as controls (n = 46). We hypothesized that fractional anisotropy (FA) would be increased acutely and partially recovered by one month post-concussion. Mood symptoms were assessed using structured interviews. FA differences were assessed using both traditional and subject-specific analyses. An exploratory analysis of tau plasma levels was conducted in a subset of participants. Results indicated that mood symptoms improved over time post-concussion, but remained elevated at T3 relative to HA. Across both group and subject-specific analyses, concussed athletes exhibited increased FA in several white matter tracts at each visit post-concussion with no longitudinal evidence of recovery. Increased FA at T1 and T3 was significantly associated with an independent, real-world outcome measure for return-to-play. Finally, we observed a nonsignificant trend for reduced tau in plasma of concussed athletes at T1 relative to HA, with tau significantly increasing by T2. These results suggest white matter abnormalities following SRC may persist beyond one month and have potential as an objective biomarker for concussion outcome. Hum Brain Mapp, 2015. © 2015 Wiley Periodicals, Inc.
[Show description] [Hide description]
- "Further development of FATCAT will include approaches for utilizing along-tract statistics in characterizing individuals' white matter properties. Also, in order to be able to account for missing data in group tables, future versions will utilize linear mixed effects (LME) modeling using AFNI's 3dLME [Chen et al. 2013]. It is also expected that, as has already happened since the initial FATCAT release, further analysis tools will be developed in response to users' requests. "
DESCRIPTION: Updates and examples of combining FATCAT, SUMA and AFNI, including: a new "mini-probabilistic" approach to tractography (as an improvement to the standard deterministic methodology); descriptions of new user-interactive visualization tools, particularly for functional and structural network connectivity, combining AFNI and SUMA; and approaches for performing group analysis of FMRI/DTI networks using 3dMVM and FATCAT command line tools.
Data provided are for informational purposes only. Although carefully collected, accuracy cannot be guaranteed. The impact factor represents a rough estimation of the journal's impact factor and does not reflect the actual current impact factor. Publisher conditions are provided by RoMEO. Differing provisions from the publisher's actual policy or licence agreement may be applicable.