fc_compute_ab_corr_kca#

qx_mri.fc.fc_compute_ab_corr_kca(flist, smask, tmask, nc, mask, root, options, dmeasure, nrep, verbose)#

fc_compute_ab_corr_kca(flist, smask, tmask, nc, mask, root, options, dmeasure, nrep, verbose)

Segments the voxels in smask based on their connectivity pattern with tmask voxels. Uses k-means to group voxels in smask.

Parameters

--flist (str):

A file list with information on sessions bold runs and segmentation files, or a well strucutured string (see general_read_file_list).

--smask (str):

Path to .names file for source mask definition.

--tmask (str):

Path to .names file for target mask roi definition.

--nc (cell array):

List of the number(s) of clusters to compute k-means on.

--mask (int | logical | vector, default 0):

Either number of frames to omit or a mask of frames to use.

--root (str, default 'flist'):

The root of the filename where results are to be saved.

--options (str, default 'g'):

A string with:

  • 'g' - save results based on group average correlations

  • 'i' - save individual sessions' results.

--dmeasure (str, default 'correlation'):

Distance measure to used.

--nrep (int, defaut 10):

Number of replications to run.

--verbose (str, default 'none'):

How to report the progress: 'full', 'script' or 'none'.

Output files

If group correlations are selected, the resulting files are:

  • <root>_group_k[N]

    Group based cluster assignments for k=N.

  • <root>_group_k[N]_cent

    Group based centroids for k=N.

If individual correlations are selected, the resulting files are:

  • <root>_<session id>_group_k[N]

    Individual's cluster assignments for k=N.

  • <root>_<session id>_group_k[N]_cent

    Individual's centroids for k=N.

If root is not specified, it is taken to be the root of the flist.

Notes

Use the function to cluster source voxels (specified by smask) based on their correlation pattern with target voxels (specified by tmask). The clustering is computed using k-means for the number of clusters specified in the nc parameter. If more than one value is specfied, a solution will be computed for each value.

Correlations are computed using the img_compute_ab_correlation gmri method. Clustering is computed using kmeans function with dmeasure as distance measure, and taking the best of nrep replications.

Examples

qunex fc_compute_ab_corr_kca \
    --flist='study.list' \
    --smask='thalamus.names' \
    --tmask='PFC.names' \
    --nc='[3:9]' \
    --mask=0 \
    --root='Th-PFC' \
    --options=g \
    --dmeasure=correlations \
    --nrep=15