On 11/14/2018 04:00 PM, Yixin Ma wrote:

External Email - Use Caution

Hi Freesurfer experts,

I'm writing to ask about coregistration between a thin oblique coronal slab of DTI image ( only 15 slices) and a cubic 1x1x1 voxel size anatomical image (volume size is 256x256x256), with additional whole brain low-resolution DTI to assist registration. The codes I use are as attached:

I have few concerns:

1. the input oblique coronal slab of  DWI image does not have

header information, should it be better if I input the DWI slab in RAS coordinates (axial volume)?

This is going to be hard to do if you don't have accurate header info. If you can get it, then the next question is whether the DWI and the anatomical were acquired at the same time. If so, then you should be able to use bbregister with --init-header. If they were acquired at separate times, is there another whole-brain image that was acquired at the same time as the DWI (any whole brain will do). If you do not have these and do not have header geometry, then you will have to set the initial registration by hand -- not very fun :(.

2. Should I zero-fill the DWI slab to match the volume of T1.mgz? I

already did that because it won't start to calculate the cost function between the thin DWI slab and anatomical image if I don't zero-fill. However, I feel like the cost function cannot make sense of the fact that the zeros are already outside the region of interest.

No.

3. Should I use --init-fsl /--init-spm in this case? Neither of these

two worked. Is there a better optimization function recommended?

No, these will fail on a slab.

I have been using fsl flirt function  2 step registration (from DWI slab to whole brain DWI; From whole brain DWI to whole brain anatomical image) to do the registration. But I don't think it's the best way.

For freesurfer bbregister, I have been trying to change things here and there, but the co-registration does not work well. Do you suggest to look at any parameter particularly based on this scenario?

Thank you for your help,

Yixin

Note that isoDWI111_256_axial.nii is the slab of DWI image EPI_iso111_256_axial.nii is the whole brain DWI image [The freesurfer reconstruction is performed before all these]

bbregister --s 30426 --mov isoDWI111_256_axial.nii --*int* EPI_iso111_256_axial.nii --init-fsl --reg xform_freesurfer.dat --dti

tkregister2 --reg xform_freesurfer.dat --mov isoDWI111_axial.nii

mri_vol2vol --reg xform_freesurfer.dat --mov isoDWI111_axial.nii --fstarg --o isoDWI111-*in*-anat.mgh

mri_convert isoDWI111-*in*-anat.mgh isoDWI111_reg.nii

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer