Hi Suresh,
yes, there is a nonlinear volumetric registration that we compute described in the 2004 paper. We also do a series of intensity normalizations, which we also describe somewhat in the 2004 paper i think. The morph has a standard max posterior term, and also a "density-based" term that really helps the medial temporal lobe align (which we also describe in the 2004 paper). You are welcome to the code if you want.
cheers, Bruce
On Tue, 9 Feb 2010, Suresh Joshi wrote:
Hello all, short time reader, first time poster...
I have a question regarding the current FreeSurfer volumetric pipeline. This page: http://surfer.nmr.mgh.harvard.edu/fswiki/FreeSurferAnalysisPipelineOverview
states that FreeSurfer has 5 stages and they are described in Fischl 2002,2004 (which I have read).
As well, there is a "a high dimensional nonlinear volumetric alignment" (which I take to be deformable registration).
So, now my question is: Can anyone give me any specifics about this non-linear transform? In the Fischl paper, the second transform gets reduced to a 12 parameter affine transform.
As well, here is the new pipeline: http://surfer.nmr.mgh.harvard.edu/fswiki/ReconAllBlockDiagram
And it doesn't clearly answer my question.
As well, there are talks of intensity normalization (ie. elimination of bias fields), but it there ever an intensity normalization step performed between the atlas and the target???
Thanks for the help, -SJ _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer