I ran CONN on my dataset but it did poorly on structural to functional registration.
I tried fsl feat gui, flirt, fnirt and mri_robust_register (freesurfer) but each of them had its specific issues.
I need to mention that I passed the functional image that was averaged over time to mri_robust_register as it do not get 4D data and I ran it with the affine parameter to do 12 DOF
Does anyone know any better way?
Hard to say what you can try without knowing precisely what you tried and seeing screenshots. You can try preprocessing in fMRIPrep, where you don’t have to worry about defining inputs incorrectly (but still more control over registration parameters), and then CONN allows you to import from fmriprep outputs.
Unfortunately my dataset is not open source so I can not post any screenshots
To make it more clear, all the results of conn is fine
The only thing I need to do is to register the skull stripped structural images to processed functional images.
Here is the codes I used so far
flirt -in skull-stripped.nii.gz -ref swaufunc.nii -out anat-registered.nii.gz
mri_robust_register --mov skull-stripped.nii.gz --dst swaufunc-avaeraged.nii.gz --lta m2d.lta --mapmov reg-anat.nii.gz --satit --affine
Hi @maryam.riazi ,
One suggestion from FSL is to use the highest resolution image as a target for
flirt. Doing so, you would need to save the transformation (
-omat argument in
flirt), invert this transformation (with
convert_xfm), and apply the inverted transformation (with
flirt -applyxfm -init). You may change the cost function as well, that could help for multimodal registration. Even better, you could use BBR to register your functional image to your anatomical image, if you have enough contrast between GM and WM in your functional image.
More explanation here:
Another option is to use
Also if you anatomical and functional images are difficult to register due to poor contrast, very different orientations between the two images, NiftyReg (reg_aladin) is a very powerful tool.
And there are many other tools from other packages: SPM, AntS, AFNI …