[MINC-users] mritotal and nu_correct

[Steve M. Robbins]

On Wed, Mar 09, 2005 at 08:23:28AM +0100, schwarzd wrote:
> Dear minc-users,
> 
> is there any comparison of registration results (accuracy) when:
> 1. mritotal is performed on native mri image
> 2. mritotal is performed on corrected mri image with N3 algorithm (nu_correct).
> 
> In my opinion, the results may be better for corrected images when thinking
> about cross correlation and mutual information. However, mritotal on native mri
> image is recommended on the web-page
> http://www.imaging.robarts.ca/Software/mni_software/mri_segment/  .
> 
> Any suggestions, explanations, links or comments?

I believe that John Sled's MSc thesis (on the N3 algorithm) has
some experiments of that nature.  Maybe the journal paper does
as well.  

-Steve