Hi Evan

it shouldn't do any anisotropic stretching. We have historical code that assumes the volume has the same width/height/depth that we haven't gotten around to getting rid of. Typically we advise people to get a 1mm isotropic volume fo reconstruction though. It takes very little time these days, and is *much* easier to analyze than 0.7mm. Also, definitely avoid analyze format as an intermediate step! Once you go through analyze it's very difficult to determine left from right.

cheers Bruce

On Thu, 21 Apr 2011, Evan Luther wrote:

Hello FreeSurfer community,

When using mri_convert to convert a .img file with a voxel size of .7 mm^3 and 252x300x300 dimensions to a nifti file for further FreeSurfer processing, mri_convert uses trilinear imterpolation to force the image to have 301x301x301 dimensions. Can anyone explain why this occurs and/or if the brain is actually being stretched in one dimension or if the image is simply being padded? Also why does it become an isotropic volume of 301 voxels rather than 300? Furthermore, once the ribbon files have been created they appear to be oriented about an origin that is at the center of the field of view. Are the surface files produced in the same coordinate system?

Thank you!