External Email - Use Caution
Thanks for your reply Douglas,
I am trying co-register the Diffusion data onto the original structural MRIs. I have tried bbregister and mri_vol2vol but the results are not great. I realized that the diffusion data have a lot of distortions, so the rigid and/or affine co-registration is not sufficient. Do you recommend performing a non-rigid registration in this case? If you do, do you know of any ways to do that within Freesurfer?
Thanks -Hedyeh ------------------------------------------------------------ ----------------------------------------------------------------- Date: Tue, 17 Apr 2018 11:49:04 -0400 From: Douglas Greve dgreve@mgh.harvard.edu Subject: Re: [Freesurfer] Non-rigid registration approaches To: freesurfer@nmr.mgh.harvard.edu Message-ID: 763da2ef-359a-9f44-9b54-1f96acb3b2ef@mgh.harvard.edu Content-Type: text/plain; charset="utf-8"
Not really. Are you trying to account for B0 distortion in the DTI? If you have a B0 map, you can pass that to bbregsiter and mri_vol2vol to account for it
On 4/16/18 10:29 AM, Hedyeh Javaheri wrote:
Hello all,
I am performing DTI analysis and wanted to know if Freesurfer has a tool for non-rigid registration. I performed recon-all and dt_recon on my subject and then used the register.dat file that was generated by dt_recon to convert the dt_recon output files back into the space of my orig.nii file from the recon-all output. Now I need to register the files in the orig.nii space back onto the T1 MRI that is in native space.
I have tried using? mri_vol2vol to get the files in the orig.nii (recon-all) space into the T1 native space without using a registration file (I used --regheader) and now the registration matrix of my file is the same as the native space, but when I use rview to visualize the registration there are flaws. I've also tried this co-registration using tkregister2 to create a registration file called register.native.dat, and then used mri_vol2vol with register.native.dat; and the results were the same as when I used mri_vol2vol with --regheader.
Is there a tool to do actual manual registration in Freesurfer? Using tkregister2 I cannot perform localized changes; and the global changes it is capable of do not improve on the coregistration.
Here is a breakdown of what I have done:
subject in Native space to Freesurfer conformed space using recon-all, output: orig.nii subject in Native space to other Freesurfer space using dt_recon, output: lowb.nii and register.dat From the other Freesufer space to Freesurfer conformed space using register.dat, output: lowb2orig.nii attempting to convert from Freesurfer conformed space to Native space: ? ? ? ? ? ? ? ? ? using mri_vol2vol and --regheader --> poor registration ? ? ? ? ? ? ? ? ? using tkregister2 to make register.native.dat, then mri_vol2vol and ?register.native.dat --> poor registration
I am running Freesurfer version 6 on Windows.
Thanks in advance! -Hedyeh Javaheri