Notes about Scale and MNI

Owned by OSAbrainresearcher1
Last updated: Oct 20, 2015 by Paul Macey (admin)
2 min read

From FSL user guide:

  • -vertexAnalysis : Set mode of operation such that vertex-wise stats are calculated. 

  • --usebvars : Set mode of operation such that it uses the combined mode parameters across the group (this is compulsory for vertex analysis). 

  • -i : concatenated bvars file containing mode parameters from all subjects (created by concat_bvars). 

  • -o : Base name of output meshes. 

  • -d : FSL design matrix (as created by the Glm GUI). 

  • --useReconNative : Reconstructs the meshes in the native space of the image. For vertex-wise stats need to also use --useRigidAlign. 

  • --useReconMNI : Reconstructs the meshes in MNI space (native space of the model). This does not require the flirt matrices. 

  • --useRigidAlign : Uses a 6 Degrees Of Freedom transformation to remove pose from the meshes (see --useScale if you wish to remove size as well). All meshes are aligned to the mean shape from the shape Model. Can be used with either ReconNative or ReconMNI. 

  • --useScale : Used in conjunction with --useRigidAlign, it will remove global scalings when aligning meshes.

 

Mail list:

The --useScale option does a 7 DOF registration for the individual meshes.  That is, it takes them from the native space and removes all pose (translation and rotation) as well as isotropic scaling (with --useScale).  This will naturally be different from removing it based on the size of the head (or ICV), especially if the size of the caudate has changed.  It is also different from the MNI space version, as in the MNI space the whole brain has been used to optimise a 12 DOF registration, allowing different scaling on different axes and also skews (or shears).
  • native space - original MRI 
  • isotropic scaling - scaling that is the same in all directions

MNI space is a standard space (an average of many individuals)
and native space is the space of the subject without adjusting it
to fit the average. The main difference is that if your subjects are
different from the "average" (e.g. older) then there can be a difference
between the MNI space and native space results (as the former will
scale each image to match the average/standard space). I would
recommend the MNI space analysis in general (as it adjusts for
head size). (link)

UseScale

  • Does rigid body registration
  • Resizes each volume to the other - accounts for structure size, not head size
    • Picks up differences in shape but not volume
  • 7 DOF - Rigid Align (6 DOF) + isotropic scaling (1 DOF)

Rigid Align

  • 6 DOF
  • Places the structures in the same place

MNI

  • Doesn't change the shape
  • Allows us to look at the volume based on the volume of head
  • Doesn't make sense to useScale and ReconMNI because MNI distorts the shape and then you're trying to look at the shape
doMVGLM

This option is used for the old-style vertex analysis. It is not needed in the new-style vertex analysis which uses randomise. (link)

 

Shape differences - ReconNative, RigidAlign, useScale

Volume differences accounting for head size - ReconMNI, RigidAlign

Volume differences - ReconNative, RigidAlign

 

 

 

Â