search

sammba-mri / sammba-mri

small mammals brain MRI
Other
16 stars 10 forks source link

affine difference for functional images after registration.fmri_sessions_to_template #70

Open garincle opened 6 years ago

garincle commented 6 years ago

Affine difference for functional images after registration.fmri_sessions_to_template on a brain 100µm template and a func_resolution at 200µm given. Affine diff near 10*10^7 image

The differance could be du to numeric error. To fixe the problem: New template at the resolution wanted only for the func_to_template function ?

salma1601 commented 6 years ago

Yes, you are right. I think AFNI can take a path instead of a resolution value to choose the grid of the output of 3dNwarpApply, so may be this would be a better option in PR #68

salma1601 commented 6 years ago

actually I was wondering if we can directly give the downsampled version of the template as master for the 3dNwarpApply. @nadkarni-na if all the registration steps are done with say dorr at 0.1mm and then we use 3dNwarpApply with master being dorr at 0.2 mm, will this be handled OK by AFNI ?

nadkarni-na commented 6 years ago

I think so

salma1601 commented 6 years ago

actually I don't know how your template200 has been generated from the template100. If I do a simple AFNI 3dresample, I get the following affine

array([[ 0.2       , -0.        , -0.        , -6.18499994],
       [-0.        ,  0.2       , -0.        , -8.08499908],
       [ 0.        ,  0.        ,  0.2       , -4.08499956],
       [ 0.        ,  0.        ,  0.        ,  1.        ]])

and this is exactly (no numeric errors) the same affine you get for your registered functional to template100 with voxel_size=(.2, .2, .2). So you need to get the information how template200 has been generated.