Hi Jahng,
> I acquired 3D T1 MPR image with 1x1x1.5(thickness) mm^3 of spatial
> resolution and DTI data with 2.4x2.4x5 mm^3.
> What is the best voxel size during writing the normalized T1 MPR images,
> which is normalized to SPM T1 template?
> I think I can choose one of 2x2x2 or 1.5x1.5x1.5 voxel size. I will segment
> the normalized T1 MPR images and apply to DTI data to mask out CSF in DTI
> data.
I am not sure there is a conclusive answer to this question. Let me see if I
get you right. You want to spatailly normalise your T1 and DTI data, segment
the T1 data and used the CSF segment to mask out CSF from the DTI image in
normalised space?
There are of course practical considerations (such as larger matrices taking
more disc space) but I think on the whole that the segementation benefits from
keeping the high resolution. Around the ventricles one sometimes see tissue
being misclassified as gray matter. This is because a voxel with a mixture of
CSF and white matter will have an intensity similar to that of a gray matter
voxel. The higher resolution you have your data in, the fewer voxels will there
be with a mixture of tissue types.
I would say that either choice (2 or 1.5) would do quite fine.
Remember to coregister your DT map to your T1 image and select it as other when
spatially normalising.
Good luck Jesper
|