I am not sure what you mean here. If one "just" resamples the images
(as fnirt does) then concentration will be preserved.
The distinction is between "preserving concentration", which is done
by just interpolating the existing values in the --in image. However,
since fnirt performs a non-linear warp it can change local volumes
(i.e. a 1cm^3 volume in the --in volume may be mapped/converted into a
2cm^3 volume in the finished resampled image). That means that if one
has "preserved concentration", i.e. just interpolated the original
values, one has effectively doubled the "amount" of whatever ones
values represents (same concentration, but double the volume).
This is a problem for some applications, notably for VBM style
analyses where it is _not_ desirable to suddenly double the local
amount of gray matter. Therefore one performs a post-processing step
where one modulates the resampled image (that has preserved
concentration) with the Jacobian-map (which is a map of local volume
changes) such that the final result does _not_ preserve local
"concentration", but rater local "amount".
I hope that clarifies things?
Good luck Jesper
On 20 Mar 2009, at 15:21, Mahinda Yogarajah wrote:
> Dear FSL Users,
> Is there an option in FNIRT normalisation equivalent to the preserve
> concentrations option in SPM5 when one writes out normalised images ?