Uses fastICA to perform spatial ICA on fMRI data.
fmri.sICA(data, mask=NULL, ncomp=20,
alg.typ=c("parallel","deflation"), fun=c("logcosh","exp"),
alpha=1, detrend=TRUE, degree=2, nuisance= NULL, ssmooth=TRUE,
tsmooth=TRUE, bwt=4, bws=8, unit=c("FWHM","SD"))object of class ''fmriICA''
list with components
4D array with ICA component images. Last index varies over components.
pre-processed data matrix
pre-processed data matrix
estimated un-mixing matrix
estimated mixing matrix
Brain mask
voxelsize
Repetition Time (TR)
fMRI dataset of class ''fmridata''
Brain mask, if NULL then data$mask is used.
Number of ICA components to compute.
Alg. to be used in fastICA.
Test functions to be used in fastICA.
Scale parameter in test functions, see fastICA.
Trend removal (polynomial)
degree of polynomial trend
Matrix of additional nuisance parameters to regress against.
Should spatial smoothing be used for variance reduction
Should temporal smoothing be be applied
Bandwidth for spatial Gaussian kernel
Bandwidth for temporal Gaussian kernel
Unit of bandwidth, either standard deviation (SD) of Full Width Half Maximum (FWHM).
Joerg Polzehl polzehl@wias-berlin.de
If specified polynomial trends and effects due to nuisance parameters, e.g.,
motion parameters, are removed. If smooth==TRUE the resulting residual series is
spatially smoothed using a Gaussian kernel with specified bandwidth.
ICA components are the estimated using fastICA based on data within brain mask.
The components of the result are related as XKW=scomp[mask,] and X=scomp[mask,]*A.
plot.fmriICA,ICAfingerprint, fastICA