|
| |
This page contains links to software programs and
components developed at Functional Neuroimaging Lab, and to other tools we've found useful and collected here for
your convenience.
Software
-
Customized,
Optimized, Automated, mostly-SPM-based MATLAB/C/Shell batch programs
performing brain imaging data processing and visualization procedures (FNLproc)
-
Automated
reconstruction programs that perform various inhomogeneity corrections,
slice-timing correction, exaction of physiological noise components from
raw functional image series, image maneuver routines such as reslicing,
re-orientation, etc.
-
Automated
functional and structural brain image preparation routines that transform
the reconstructed raw images into specific data format and data structure,
ready for further image processing and analysis steps.
-
Automated
and optimized image processing procedures, algorithms and programs for
both functional and structural brain images, including:
-
Correction
for geometric distortion due to magnetic field inhomogeneities based
on the B0 field image;
-
Realignment
of functional image series to correct for slight head movement between
scans based on intracranial voxels;
-
Co-registration
of functional images to the corresponding high-resolution anatomical
image based on the rigid body transformation parameters of the
reference anatomical image (with the same axial slice placement and
thickness as the functional imaging) to the latter for each individual
subject;
-
Stereotactic
normalization to a standardized coordinate space (Montreal MRI Atlas
version of Talairach space) based on the intracranial voxels of the
high-resolution anatomical image to normalize for individual
differences in brain morphology;
-
Spatial
smoothing with an isotropic Gaussian kernel.
-
Customized,
Optimized, Automated, mostly-SPM-based MATLAB/C batch programs
performing standard statistical analyses and related procedures (FNLstat)
-
Automated and optimized routine that define the brain region in
functional images for each subject, the temporal global fluctuation is
then estimated as the mean intensity within brain region of each volume;
-
Automated
GLM routines for building voxel-by-voxel univariate multiple linear
regression model at the subject level to determine the extent to which
each voxel's BOLD activity correlated with the principal regressor;
-
Automated
programs for group-level random-effects analyses including paired
t-test, two-sample t-test, single/multiple regression, ANOVA and ANCOVA;
-
Automated
and optimized VBM (voxel-based morphometry) analysis program based on
structural MRI images;
-
Automated
programs for effective connectivity analysis: using adjusted functional
BOLD signal from a seed ROI as the principal regressor in a multiple
linear regression model for each subject, then at the group level
employing a random-effects model to summarize the within-group
correlation levels and the between-group differential correlation levels
of the other brain regions with the seed ROI;
-
Interactive
GUI interface/automated routines performing statistical inference such
as two-tailed t-tests with corresponding statistical summary tables,
multiple options of brain activation rendering layout and automated
printing programs.
-
Customized,
Optimized, Automated fmristat-based MATLAB/C batch programs performing
two-stage linear mixed-effects modeling of fMRI data (FNLlme)
-
MATLAB
batch programs performing two-stage voxel-wise linear mixed-effects models
of fMRI data, which are able to utilize the same set of voxel-wise GLM
from prototypical analysis, connectivity analysis to VBM, with reduced
biases in effect estimation and more proper modeling of the residuals;
-
Statistical diagnostic
programs using partial correlation method to assess the relative
contributions of variance components in a multiple regression model;
-
Streamlined programs
utilizing high-performance computing clusters to increase computational
feasibility of advanced statistical modeling and evaluation procedures at
least by a factor of 50-100.
-
Advanced
Statistical Data Analysis Methods Development
-
Multi-level linear/nonlinear mixed-effects models for nested factors
in fMRI data, which address the heteroscedasticity in variance-covariance
structure and further reduce the biases in effect estimation (the programs
are R-and-C-based);
-
New
generation of R/C/Shell based computer programs utilizing high-performance
computing clusters to overcome various disadvantages of software tools
currently available in neuroimaging field (conventional approaches have
primarily relied on various oversimplifications and less-than-optimal
methods in favor of low computational cost, recent studies have shown that
such a practice can lead to severely biased estimations).
-
Multivariate
Analysis Methods Development
-
Confirmatory
principal component analysis of fMRI data that can be performed within and
across groups to detect statistically significant topographic regional
patterns of brain activity, and the association between the degree to
which each subject expresses such patterns (numerated by loading scores of
each subject), and the corresponding clinical and other measures, can then
be evaluated through correlation analysis;
-
Independent
component analysis of fMRI data based on nonparametric density estimation;
-
Statistical
Graphic Model/Structure Equation Modeling/Path Analysis for testing
hypotheses concerning the group and condition differences in correlations
among the specified brain regions.
-
User-programmable
Database System (FNLBD)
Tools
|
