Using Workbench Command

Workbench Command is a set of command-line tools that can be used to perform simple and complex operations within Connectome Workbench.

Full Documentation:

Documentation Home
   wb_command -cifti-parcellate
      <cifti-in> - the cifti file to parcellate
      <cifti-label> - a cifti label file to use for the parcellation
      <direction> - which mapping to parcellate (integer, ROW, or COLUMN)
      <cifti-out> - output - output cifti file

      [-spatial-weights] - use voxel volume and either vertex areas or metric
         files as weights

         [-left-area-surf] - use a surface for left vertex areas
            <left-surf> - the left surface to use, areas are in mm^2

         [-right-area-surf] - use a surface for right vertex areas
            <right-surf> - the right surface to use, areas are in mm^2

         [-cerebellum-area-surf] - use a surface for cerebellum vertex areas
            <cerebellum-surf> - the cerebellum surface to use, areas are in

         [-left-area-metric] - use a metric file for left vertex weights
            <left-metric> - metric file containing left vertex weights

         [-right-area-metric] - use a metric file for right vertex weights
            <right-metric> - metric file containing right vertex weights

         [-cerebellum-area-metric] - use a metric file for cerebellum vertex
            <cerebellum-metric> - metric file containing cerebellum vertex

      [-cifti-weights] - use a cifti file containing weights
         <weight-cifti> - the weights to use, as a cifti file

      [-method] - specify method of parcellation (default MEAN, or MODE if
         label data)
         <method> - the method to use to assign parcel values from the values
            of member brainordinates

      [-exclude-outliers] - exclude non-numeric values and outliers from each
         parcel by standard deviation
         <sigma-below> - number of standard deviations below the mean to
         <sigma-above> - number of standard deviations above the mean to

      [-only-numeric] - exclude non-numeric values

      [-fill-value] - specify value to use in empty parcels (default 0)
         <value> - the value to fill empty parcels with

      [-nonempty-mask-out] - output a matching pscalar file that has 0s in
         empty parcels, and 1s elsewhere
         <mask-out> - output - the output mask file

      [-legacy-mode] - use the old behavior, parcels are defined by the
         intersection between labels and valid data, and empty parcels are

      [-include-empty] - deprecated: now the default behavior

      Each label (other than the unlabeled key) in the cifti label file will be
      treated as a parcel, and all rows or columns of data within the parcel
      are averaged together to form the parcel's output row or column.  If
      -legacy-mode is specified, parcels will be defined as the overlap between
      a label and the data, with no errors for missing data vertices or voxels,
      and empty parcels discarded.  The direction can be either an integer
      starting from 1, or the strings 'ROW' or 'COLUMN'.  For dtseries or
      dscalar, use COLUMN.  If you are parcellating a dconn in both directions,
      parcellating by ROW first will use much less memory.

      NOTE: the parcels in the output file are sorted by the numeric label
      keys, in ascending order.

      The parameter to the -method option must be one of the following:

      MAX: the maximum value
      MIN: the minimum value
      INDEXMAX: the 1-based index of the maximum value
      INDEXMIN: the 1-based index of the minimum value
      SUM: add all values
      PRODUCT: multiply all values
      MEAN: the mean of the data
      STDEV: the standard deviation (N denominator)
      SAMPSTDEV: the sample standard deviation (N-1 denominator)
      VARIANCE: the variance of the data
      TSNR: mean divided by sample standard deviation (N-1 denominator)
      COV: sample standard deviation (N-1 denominator) divided by mean
      L2NORM: square root of sum of squares
      MEDIAN: the median of the data
      MODE: the mode of the data
      COUNT_NONZERO: the number of nonzero elements in the data

      The -*-weights options are mutually exclusive and may only be used with
      MEAN (default), SUM, STDEV, SAMPSTDEV, VARIANCE, MEDIAN, or MODE (default
      for label data).