https://github.com/nilotpalsanyal/bhmsmafmri
R package for Multi-Subject Multiscale Analysis of fMRI Data, or Other Multiscale Data
https://github.com/nilotpalsanyal/bhmsmafmri
bayesian-hierarchical-models fmri-data-analysis multiscale-data wavelet-transform
Last synced: 2 months ago
JSON representation
R package for Multi-Subject Multiscale Analysis of fMRI Data, or Other Multiscale Data
- Host: GitHub
- URL: https://github.com/nilotpalsanyal/bhmsmafmri
- Owner: nilotpalsanyal
- Created: 2022-09-05T20:05:51.000Z (about 4 years ago)
- Default Branch: main
- Last Pushed: 2024-06-13T02:07:03.000Z (over 2 years ago)
- Last Synced: 2025-09-23T15:56:32.079Z (12 months ago)
- Topics: bayesian-hierarchical-models, fmri-data-analysis, multiscale-data, wavelet-transform
- Language: C++
- Homepage: https://nilotpalsanyal.github.io/BHMSMAfMRI/
- Size: 2.17 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# BHMSMAfMRI: Bayesian Hierarchical Multi-Subject Multiscale Analysis of fMRI Data
[](https://CRAN.R-project.org/package=BHMSMAfMRI)
[](https://github.com/nilotpalsanyal/BHMSMAfMRI/actions/workflows/R-CMD-check.yaml)
[](https://www.codefactor.io/repository/github/nilotpalsanyal/bhmsmafmri)
[](https://cran.r-project.org/package=BHMSMAfMRI)
BHMSMAfMRI performs Bayesian hierarchical multi-subject multiscale
analysis of function MRI (fMRI) data as described in Sanyal & Ferreira
([2012](#ref-paper)), or other multiscale data, using wavelet based
prior that borrows strength across subjects and provides posterior
smooth estimates of the effect sizes and samples from their posterior
distribution.
## Installation
### Install from CRAN
``` r
install.packages("BHMSMAfMRI")
```
### Install from GitHub
``` r
# install.packages("devtools")
devtools::install_github("nilotpalsanyal/BHMSMAfMRI")
```
## The main function:
BHMSMA is the main function which accepts fMRI data as a 4D array (see
code below) and a design matrix. For the time-series of all voxels, a
general linear model (GLM) is fit with all the regressors in the design
matrix. After that, the standardized regression coefficient map of a
regressor of interest is subjected to further analysis. The function
BHMSMA returns the posterior smoothed map of the regression
coefficients. Below is a basic illustration of its use. For a detailed
manual, see the package
vignette.
``` r
library(BHMSMAfMRI)
#>
#> Welcome! Thanks for trying BHMSMAfMRI.
#>
#> Website: https://nilotpalsanyal.github.io/BHMSMAfMRI/
#> Bug report: https://github.com/nilotpalsanyal/BHMSMAfMRI/issues
# Read data from image files
fpath <- system.file("extdata", package="BHMSMAfMRI")
untar(paste0(fpath,"/fmridata.tar"), exdir=tempdir())
n <- 3
grid <- 32
ntime <- 9
data <- array(dim=c(n,grid,grid,ntime))
for(subject in 1:n)
{
directory <- paste0(tempdir(),"/fmridata","/s0",subject,"/")
a <- readfmridata(directory, format="Analyze", prefix=paste0("s0",subject,"_t"),
nimages=9, dim.image=c(grid,grid,1))
data[subject,,,] <- a[,,1,]
}
data(fmridata)
names(fmridata)
#> [1] "grid" "nsubject" "TrueCoeff" "DesignMatrix"
truecoef <- fmridata$TrueCoeff
designmat <- fmridata$DesignMatrix
# Perform analyses
k <- 2 #consider the second regressor
analysis <- "multi" #perform multi-subject analysis (MSA)
BHMSMAmulti <- BHMSMA(n, grid, data, designmat, k, analysis, truecoef)
analysis <- "single" #perform single subject analysis (SSA)
BHMSMAsingle <- BHMSMA(n, grid, data, designmat, k, analysis, truecoef)
# Compare results for the first subject
zlim = c(0,max(abs(BHMSMAmulti$GLMCoefStandardized[1,,,k])))
par(mfrow=c(2,2))
image( truecoef[1,,],col=heat.colors(12),main="true map")
image( abs(BHMSMAsingle$GLMCoefStandardized[1,,,k]),
col=heat.colors(8),zlim=zlim,main="GLM coef map")
image( abs(BHMSMAsingle$GLMcoefposterior[1,,]),
col=heat.colors(8),zlim=zlim,main="posterior map SSA")
image( abs(BHMSMAmulti$GLMcoefposterior[1,,]),
col=heat.colors(8),zlim=zlim,main="posterior map MSA")
```
## References:
Sanyal, Nilotpal, and Ferreira, Marco A.R. (2012). Bayesian hierarchical
multi-subject multiscale analysis of functional MRI data. Neuroimage,
63, 3, 1519-1531.