Hi,
It is not necessary to do any intensity normalisation for either segmentation or registration, as all of our tools are either invariant to scaling or estimate the relative values directly from the image.
All the best,
Mark
On 15 Feb 2013, at 20:46, Vincent Koppelmans <[log in to unmask]>
wrote:
> Hi all,
>
> I noticed that some automated pipelines from other software packages apply intensity normalization (adjusting the global intensity range). For example, FreeSurfer scales the intensities for all voxels so that the mean intensity of the white matter is 110. BRAINS rescales images so that an intensity of 10 represents pure cerebrospinal fluid, 130 pure gray matter, and 250 indicates pure white matter.
>
> I know that FSL FEAT does grand mean scaling for fMRI, but I am not aware of intensity scaling/normalization in FSL's structural tools, either optional or by default. Is that correct? If so, should I apply intensity normalization on my 3D T1 files (for example using fslmaths -inm), or is this not a necessary pre-processing step before the segmentation.
>
> Thanks, Vincent
|