vbmaskinfo

vbmaskinfo takes as input a 4D volume in which each "time" point is a lesion mask. It prints a summary of the lesion masks, including a count of the number of distinct voxels and a breakdown of the number of voxels with different lesion counts.

Here's the help text:

VoxBo vbmaskinfo (v1.8.5/Jun 14 2011)
summary: mask/lesion summary utility for VoxBo
usage:
  vbmaskinfo <flags> <img> [<img> ...]
flags:
  -m <mask>      inclusion mask (include voxels with non-zero values)
  -n <num>       minimum lesions for a voxel to be included (default 1)
  -u <file>      write unique-voxel volume
  -s <file> <n>  select n random distinct in-mask voxels, write as matrix
  -r             see below
  -h             show help
  -v             show version
notes:
  vbmaskinfo treats all the passed 3D and 4D volumes as one
  large series of masks, typically lesion masks, and produces
  a summary of the number of total and distinct voxels in the
  combined volume.

  The -u option creates a volume with only unique voxels.  For every
  set of n identical voxels (voxels that have the same value at every
  point in the series), only one will be included.

  The -r flag is used to extract matrices of feature vectors
  heuristically from a 4D mask set.  Each feature vector contains the
  data from a single voxel, as a column in the matrix.  E.g., if you
  have 20 masks and you extract 50 features, your feature matrix will
  be 20x50.  The -r flag must be followed by five arguments, as
  follows:

    -r <mfile> <map> <nfeatures> <maxrmul> <countweight>

  mfile is the name of the matrix output file (something.mtx for a
  text matrix).  map is the name of the 3D file to be used as output
  (use "none" to suppress).  nvoxels is the maximum number of features
  you'd like to extract.  maxrmul is the maximum R value for a model
  of the candidate vector, with the already-included vectors as the
  IVs (the higher the maxrmul, the less the multicollinearity among
  the features).  if maxrmul is 1.0 or greater, then all voxels will
  be included (this is useful if you want to create a "district" map
  of identical voxels).  countweight is the amount of weight given to
  the number of voxels containing the pattern to be considered.
  countweight can be given either in the 0-1 range or in the 0-100
  range.  the remaining weight is given to the feature's variance.  if
  both weighting parameters are non-zero, then the numbers used are
  scaled to the 0-1 range, so that they're at least roughly
  comparable.

  Note that when writing matrix files using -s or -r, the .mat
  extension will get you a VoxBo matrix (MAT1) file, while the .mtx
  extension will get you a text format matrix file with features in
  columns and subjects (volumes) in rows.  Lines beginning with # are
  comments.