Dear Tortoise Team,
Thanks for your wonderful support and your new fast and awesome Tortoise 3.1.4!
Having read in a previous Tortoise manual that one should use a fat saturated T2 weighted structural image to perform a proper EPI distortion correction between the structural and the B0 we could still register our B0 images very well with our plain (no fat sat.) T2 weighted images in Tortoise 2.5.2b (IDL version), using the “ITK_structural_registration parameter” set to “on” in the active dmc-file for prepvm. Using the plain T2 weighted was not 100% in Tortoise 3.1.4 since a few subjects got some kind of leak problem. However, we seem to have found two workarounds and wonder which you feel is best.
1) We believe that the biggest difference between a fat saturated T2 weighted image and a plain T2 weighted image is that the skin outside the skull is very bright in the plain T2 weighted imaged. Running the FSL bet2 skullstrip-command on our raw T2 weighted image, the bright skin outside the skull was removed and our “bleeding” problems were eliminated for about a hundred tested subjects so far. In addition, the registrations between B0 and structural looks wonderful, using the “ANTSSyN” “EPI_optimization_settings” in the dmc-file we run DIFFPREP command (in Tortoise 3.1.4).
2) According to previous DIFFPREP post by Wyllians Borelli (titled “Missing clusters after DIFFPREP”, posted 03/29/2019 - 12:55) it seems as one may register B0 to the subject’s T1 weighted image after running fat_proc_imit2w_from_t1w. We have not tried this approach but believe it may work as well.
I understand that a fat saturated T2 is best, but in the case that that T2 does not exist we wonder which workaround you believe is best between method 1 or 2 above if one wants to use EPI distortion correction and one still want to run subjects in the latest Tortoise version?
It's great that the registration quality looks good despite the use of non fat suppressed T2w structural image in distortion correction. We highly recommend the use of a fat suppressed T2 w structural to reduce the influence of soft tissue around the brain stem or around the brain (at the tissue air interface) in correction of artifacts in these regions. Most of the distortions in your EPI images occur at the tissue air interfaces. If your distortion correction has performed well in these regions, then perhaps the settings are optimal to perform good correction or soft tissue signal present in the structural may be not as strong.
As far as best approach to use in the absence of a T2w fat suppressed image in the correction, you may try with available structural image you have (be it non fat suppressed structural or T1 to T2 converted structural), but be sure to check if these regions surrounded by soft tissue in the structural has not caused the corrections to not be ideal in your corrected b0 or final corrected DWI's.
Thanks for detailed answer so we further may optimize our Tortoise results!