tissue classification

Important: PVC should only be applied to skull-stripped brain images Note that the models used in PVC assume that only voxels containing CSF, grey matter, and white matter remain in the image. Thus, images should always be skull-stripped before applying PVC.

The next step in the BrainSuite Cortical Surface Extraction sequence is tissue classification, which is performed by the Partial Volume Classifier (PVC). This process assigns an integer tissue label to each voxel in the image. These labels correspond to the type of tissue that is estimated to be in that voxel. PVC accounts for background voxels, cerebrospinal fluid, grey matter, and white matter. It also labels voxels that are composed of combinations of these voxels.

The PVC algorithm uses the partial volume measurement model that was used for the bias field correction. However, it removes the spatially varying gain component since that is assumed to have been corrected by BFC. It does, however, include a spatial structural prior. This prior models the brain as a set of piecewise continuous regions of single tissue types, bounded by partial volume combinations, and it encourages neighborhoods of voxels to be similar. The influence of this prior is controlled by the Spatial Prior parameter.


  1. Tissue Classification

    The tissue classification step performs a maximum-likelihood estimate of the tissues present in the image based on the partial volume measurement model.
  2. MAP Classification

    The MAP classification step iterates to find a solution to the maximum a posteriori classifier that incorporates the partial volume measurement model with the spatial prior.


    Spatial Prior The spatial prior can be adjusted to reduce its influence, which can be useful if the image has low signal-to-noise. Larger values will produce smoother contours in the tissue classification. If the value is too large, then the regions will be too smooth and the contours will not reflect the anatomy.

Command-Line Usage

pvc: performs voxel-wise tissue classification on T1-weighted MRI. Image should be skull-stripped and bias-corrected before tissue classification.

usage: pvc -i input [optional settings]

example: pvc -i mri.nii.gz -o mri.pvc.label.nii.gz -f mri.frac.nii.gz

Required Settings:
Flags Description
-i <input MRI> MRI file
Optional Settings:
Flags Description
-m <mask file> brain mask file
-o <label file> output label file
-f <fraction file> output tissue fraction file
-l <lambda> spatial prior strength [default: 0.1]
-v <level> verbosity level (0 = silent) [default: 1]
--init <initfile> initialization file
-3 use a three-class (CSF=0,GM=1,WM=2) labeling
--timer time processing

Input MRI should be skull-stripped or a mask should be specified.
Output tissue label volume is an 8-bit label file using the values:
0 (CSF), 1 (GM), 2 (WM), 3 (GM/CSF), 4 (GM/WM), 5 (CSF/Other), 8 (Background)
Tissue fractions are saved as a 32-bit floating point volume:
0 (BKG) 1 (CSF) <-> 2 (GM) <-> 3 (WM)

Example Result

Output Files

If “save output of each stage automatically” is checked on the Cortical Surface Extraction dialog, the following files are generated (where filename_prefix is the filename of the MRI scan without the file extension, e.g. “testsubj” for the file “testsubj.nii”):

Filename Contents
filename_prefix.pvc.frac.nii.gz Label volume
filename_prefix.pvc.label.nii.gz Label volume of tissue types. Contains labels for white matter, grey matter, CSF, and combinations of each.

Restore from Previous Session

If BrainSuite was interrupted while performing this stage or to change the parameters for this stage and rerun after fully processing a scan, click “Restore From Previous…” on the bottom of this stage’s dialog box and load the original MRI scan. BrainSuite will automatically load all of the files generated in previous stages, allowing processing to restart from this intermediate stage.