Combine ICA results from multiple runs or multiple subjects in group fMRI studies
fmri.sgroupICA(icaobjlist, thresh = 0.75, minsize=2)
An object of class ''fmrigroupICA'' with components
Mean IC's over cluster members for cluster of size larger
or equal minsize
Size of selected clusters
Number of selected clusters
Cluster membership corresponding to thresh.
Distance value at which the cluster was created. Elements correspond to elements of cluster.
Object returned by function hclust.
List of results obtained by function fmri.sICA
for a series of fmri data sets (multiple runs or multiple subjects).
threshold for cluster aggregation. Needs to be in (0,1).
Minimal size of cluster to consider in IC aggregation. Needs to be larger than 1.
Joerg Polzehl polzehl@wias-berlin.de
The fMRI time series need to be preprocessed and registered before thr ICA decomposition is performed.
The function employs a hierarchical clustering algorithm (complete linkage) on the combined set of spatial independent components obtained from the individual time series. A distance matrix is obtained from correlations of the independent component images. Aggregation of two components from the same fmri series is prevented in the algorithm.
F. Esposito et al (2005) Independent component analysis of fMRI group studies by self-organizing clustering, Neuroimage, pp. 193-205.
fmri.sICA, plot.fmrigroupICA, hclust