Masks registration

[findios]

Hi There's any way i can register one image using autoreg without changing the image intensities? I'm trying register one file that contain the left and right hippocampus labels (two different labels) and after registration i can't separate them.

Thanks Patrick

[andrewjanke]

Hi Patrick,

If I understand you correctly, you'll want to add a '-nearest_neighbour' argument when you call mincresample to avoid interpolation.

a

[findios]

Thanks you Andrew for the fast reply. That was exactly what i needed. Regards

[findios]

I'm sorry for asking something like this again but: after i register one image with the standard_pipeline if I want t register the hippocampus mask of that image shouldn't i just use: mincresample -nearest_neighbour / -transformation transformation_file_from_pieline.xfm / -like mni_icbm152_t1_tal_nlin_sym_09c.mnc / mask.mnc mask12_reg.mnc

Something is going wrong with the resampling. Thanks for the help in advance. Patrick

[vfonov]

Hello,

can you define precisely what kind of data you have and what are you trying to achieve?

[findios]

Both images (MR image an manually segmented hippocampus mask image) were download from ADNI database. Basically I will need to know which potions in the original image will correspond to the hippocampus after the registration. The registration of the original image was made using standard_pipeline.pl. Now I was trying to apply the same registration to the mask file. Sorry if i'm note making myself clear enough.

[vfonov]

Then , I suggest using itk_resample --byte --labels --transform transformation_file_from_pieline.xfm --like mni_icbm152_t1_tal_nlin_sym_09c.mnc input.mnc output.mnc

Also, make sure that labels align with the anatomy as was used as input for the pipeline (i.e in Display or register )

[findios]

The masks appear to have some translation. When i use mincinfo the start point isn't the same. That could result in some problem in the registration?

[vfonov]

Check if the voxels in masks properly align with anatomical scans using register or display (i.e minc programs).

[findios]

No, they are not. The MRI was preprocessed so probably there is some kind translation I wasn't aware. I will try using the original image. Thank you so much for the help. Regards Patrick