Child pages
  • minc-stuffs (formerly mice-minc-tools)
Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 11 Next »

minc-stuffs repository

MICe-minc-tools are now part of the minc-stuffs github repository. Any updates to the code can be downloaded and installed from there.

MICe-minc-tools overview

MICe-minc-tools is a software package containing disparate command line tools operating on MINC files developed at the Mouse Imaging Centre. The tools are:


Brief Description


Create a vector volume containing the displacements at each voxel specified by a transform (.xfm file).


Compute the volume of segmented structures given an atlas and Jacobian determinants.


Create a tag file from a transform.


Compute the linear part of a non-linear transform.


Multiply values in a file by the combined scale factor of a linear transform.


Blur a vector file.


Pretty pictures of deformation fields


Add normally distributed noise to a volume.

TFCEThreshold-free cluster enhancement

Installation instructions

Get the code here

The C code relies on MINC-2.0.09 or greater installed. #smooth_vector requires python, and #add_noise_to_volume requires R and RMINC, and the rest requires perl, Getopt::Tabular, and mni_perllib.

After untarring the tarball, the following commands will perform the install:

./configure --with-minc2 --prefix=/directory/to/install/to --with-build-path=/directory/containing/minc2
make install

Description of included programs.



minc_displacement [options] input.mnc input.xfm output_grid.mnc

minc_displacement evaluates a transform - linear nonlinear, concatenated, etc - at every voxel of an input file and produces a 4D output volume containing the displacement in x,y,z at every voxel.



label_volumes_from_jacobians structures.mnc jacobians.mnc

label_volumes_from_jacobians produces a list of labels and their volumes from a segmented atlas and a volume of log jacobians. Some important notes:

  • The exponent of the input jacobians is taken - it thus has to be a volume of logs of jacobians.
  • The atlas volume and the jacobians have to be in the same space with the same sampling - i.e. evaluations take place in voxel coordinates.



xfm2tag [options] input.mnc input.xfm output_tags.tag

xfm2tag takes a transform and an input volume in the space of which the transform can be evaluated and produces a tag file similar in composition to what register would produce after manual registration. This tag file can then be used to create a newer and perhaps simpler transform using tagtoxfm. xfm2tag is the key tool behind lin_from_nlin.

An optional argument allows one to specify a mask - the transform will only be evaluated inside the mask.



lin_from_nlin [options] input_volume.mnc input_nonlin.xfm output_lin.xfm

lin_from_nlin takes a non-linear transform and an input volume over which the non-linear transform can be evaluated and computes the linear component of the non-linear transform. Options can be used to specify lsq12, lsq9, or lsq6 for the resulting linear transform.



scale_voxels [options] transform.xfm volume.mnc output.mnc

scale_voxels multiplies every voxel in an input volume by the scale factor determined from a linear transform.



smooth_vector --filter --fwhm=x input_vector.mnc output_vector.mnc

Blurs a vector volume by separating the vector components, blurring them individually using mincblur, then recombining them.



grid_object_manipulator -mincfile mincfile.mnc -transform transformation.xfm

grid_object_manipulator uses ray_trace to create pictures of what a deformation grid looks like. Lots of options to control the grid-point spacing, use of overlays, and other ray_trace options. The image below is one example generated by grid_object_manipulator (click on it for a larger image):



add_noise_to_volume [options] -noise x input.mnc output.mnc

What every lab needs - a way to make their images look worse. add_noise_to_volume adds x percent normally distributed noise to an input volume. Needs R with the RMINC library installed.

  • No labels