Dear Gabor,
> I've read that sinc interpolation is important for quantitative
> imaging, but I've not found anything about the optimal width of Sinc
> Window (full width is in voxels - voxels of input volume or
> reference volume??). Could somebody help me which window width I
> should use?
there is no such thing as an "optimal" window width. It will always be
a compromise where greater width yields better quality but longer
execution time.
I myself would opt instead for cubic spline interpolation where
increased quality comes at a much lower prize (in terms of execution
time) thanks to the existence of a recursive algorithm. You cannot use
spline interpolation directly in flirt, but you can use the affine
transformation derived from flirt as input to the --premat parameter
in applywarp and specify -interp=spline.
Good luck Jesper
|