Hi,
> I am dealing with similar issues and your posts are really helpful. I am just wondering if creating lesion mask is necessary when one is just registering a patient's functional image to his own structural image, without furthering registering to any standard space.
If both images show the lesion then you do not need a lesion mask.
> Also, would 6DOF be enough for functional to structural registration for stroke patients , or other methods such as BBR are recommended?
We strongly recommend using a fieldmap or blip-up-blip-down data whenever dealing with EPI acquisitions (typical for functional or diffusion images) due to the distortions. The best method we have of registering images that includes this information is with BBR.
If you do not have fieldmaps then we still recommend BBR (with 6 DOF) as this tends to do a better job, but we strongly recommend the use of fieldmaps (or blip-up-blip-down data) in addition to BBR.
The only problem you might have with BBR is obtaining a white matter segmentation. However, it is not crucial that the segmentation is purely white matter - the main thing that is important is whether the edge of the segmentation coincides with an intensity change in both images. If that is the case (whether it involves the lesion or not) then you can happily use it with BBR. If the intensity change is different in different parts of the image then you might benefit from using the "-bbrtype local_abs" option.
All the best,
Mark
>
> Thank you!
> Yuqi
|