Hi all
When you loaded the lowb into freeview and it looked registered with the anatomical, did you load the lowb with the registration file? If not, then the registration may be wrong. Things may look almost right when you don't specify a registration file because you probably collected the dti and anatomical in the same session. If so, then check the registration with something like
tkregister2 --mov lowb.nii.gz --reg anatorig2diff.bbr.dat --surfs
I did not load it with the registration file, but I just did and it does not change the results on freeview. I also checked the registration on tkregister and it seems very fine to me.
also, what is the first value in anatorig2diff.bbr.dat.mincost?
Here are all the values: 0.180949 209.653007 285.934948 29.944402
Maybe one point that I don't understand is why we have to apply a registration matrix if all the volumes (brainmask.mgz, lowb.nii.gz and the lesion mask) are already registered? Because if I am applying the invert of registration matrix anatorig2diff.bbr.dat to the lesion mask, then it will be in the native diffusion space, not the recons space anymore, isn't it?
doug
On 2/17/14 10:41 PM, Celine Louapre wrote:
The lesion_vol2vol volume is still not correctly overlapping the DTI.
The only command that didn't give me registration issue is mri_convert where I did not put any registration file but I was not completely happy with the interpolation resampling. (nearest missed some lesions, maybe interpolate and then binarize the mask was one possible option). I can show you the results tomorrow if you prefer Celine
what does it look like if you run
tkmedit -f lowb.nii.gz -ov lesion_WM_vol2vol.nii.gz -fminmax .5 1
On 2/17/14 10:00 PM, Celine Louapre wrote:
No problem, sorry! So when I am loading my lesion mask on freeview on brainmask.mgz and on DTI_fit_FA.nii, the lesions are correctly registered on both DTI and T1. The only thing I would like to do is to resample the lesion mask that has a resolution of 0.3*0.3*1 on the resolution of the DTI that has 1.8mm^3. When I am using the command below with mri_vol2vol, and the registration matrix anatorig2diff.bbr.dat, the output is a lesion mask that has 1.8mm3 resolution, but the lesions are not overlapping the other modalities and the previous mask on freeview. Is it the way I am loading it on freeview maybe? or actually all the volumes are already on the recons space so maybe I should not use any registration matrix? Celine
oops, sorry! That command looks like it is correct. Can you say more about what is going wrong? How are you determining that something is going wrong? doug
On 2/17/14 9:30 PM, Celine Louapre wrote:
Hi Doug I don't know if you had the chance to read my email below to resample a lesion mask on diffusion images. Anastasia gave me the registration file to use (anatorig2diff.bbr.dat) in the command but the output mask file was not registered anymore. So she asked me to rephrase my question more generally on the mailing list to look for help with bbregister. Thanks a lot Celine
---------------------------- Original Message
Subject: [Freesurfer] Resample a mask on diffusion images using bbregister registration matrix From: "Celine Louapre" celine@nmr.mgh.harvard.edu Date: Thu, February 13, 2014 9:58 pm To: freesurfer@nmr.mgh.harvard.edu
Hello freesurfer team I have binary lesions mask from T2star images (resolution 0.3*0.3*1) already registered on freesurfer space that I would like to resample on diffusion images to further exclude the lesion voxels on the DTI maps (and to calculate DTI metrics within the lesions too). Using mri_vol2vol command I think I am missing something because I tried with the registration file produced by bbregister (through tracula) but the result does not look correct. I have the lesion mask in the freesurfer space that I can overlay on brainmask or dtifit_FA.nii on freeview and it seems correctly registered on both. The command I am running is: mri_vol2vol --mov recons_dir/subj1/dmri/lowb.nii.gz --targ /lesion_mask/subj1/lesion_WM.nii.gz --interp nearest --o /lesion_mask/subj1/lesion_WM_vol2vol.nii.gz --reg recons_dir/subj1/dmri/xfms/anatorig2diff.bbr.dat --no-save-reg --inv
And then when I open lesion_WM_vol2vol.nii.gz in freeview, the voxel size is the same as the DTI images, but the lesions are not at the same place.
On the mailing list I have also seen other options but I am not sure if they would be appropriate? I have seen mri_label2vol command where I could set the fill threshold parameter but it does not solve the problem of the registration file? also mri_convert And eventually I also saw this command: mri_mask but I guess that the mask volume needs to be in the same dimension as the diffusion space?
Thanks a lot Best Celine