Script to remove small clusters or standalone voxels from an ROI/atlas dataset

Usage Example

 @ROI_decluster                                       \
   -input  mydset.nii.gz                              \
   -outdir myoutdir                                   \
   -fracthresh  0.15                                  \
   -prefix newroi_name

Note only the input dataset and a threshold are required.


    -input input_dset   :required input dataset. This dataset should be
                         set of integer values. The program mostly assumes
                         approximate isotropic voxels.

    -outdir outdirname  :directory name for output. All output goes to
                         this directory. Default is roidc.

    -nvox_thresh nn     :number of voxels in a cluster to keep
    -frac_thresh nn     :fraction of voxels in a cluster to keep [0.0-1.0]
                         Both types of threshold can be specified at the same
                         time, in which case, the minimum value would be used.
                         For example, an nvox_thresh of 10 and a frac_thresh
                         of 0.15 would remove all voxels that occupied at least
                         15% of the number of voxels in the region and at least
                         10 voxels.

    -prefix baseprefix  :base name of final output dataset, i.e. baseprefix.nii.gz
                         Default is rdc, so output would be rdc.nii.gz

    -NN [1,2,3]         :neighborhood type using in finding mode,
                         1 - facing neighbors, 2-edges, 3-corners

Also see these programs with these related functions:
   3dClusterize - reports and extracts clusters (main program called by this one)
   @ROI_modal_grow - grows regions using non-zero modal smoothing
   3dROImaker - grows regions using regular dilation iteratively