I am trying to make binary lesion masks in MRICRON for a set of T1 scans,
and then use those masks to normalize and segment T1 scans in SPM5.
I noticed that MRICRON generates a binary mask that has different dimensions
compared to the original T1.
Here are the steps I followed for one image:
1. Draw the mask on original T1: xyz dimensions 512 512 72, with z having
2mm spacing. 16bit big endian
2. Save the mask as *.voi
3.Then convert *.voi to Nifti
4.I end up with a binary mask (*.nii): xyz dimension 512 512 331 with z
having 0.4297 spacing. 8bit little endian.
Using this mask to map the lesion in SPM5 and normalizing the image would
produce an image with wrong lesion map (i.e. the mask is applied to the
wrong slices). I have tried coregistering and reslicing the lesion map in
SPM5 but I still get an image with lesion mapped incorrectly.
I would like to know how to solve this problem or if there are better ways
of making binary lesion masks.
Thanks,
Sarvy
|