Dear FSL experts,
I have some issues in DTI preprocessing which make me confused a bit. I wanted to share with you and find out if you have any idea.
For doing a DTI analysis using DTIFIT, I follow the pipeline proposed on the FSL website.
My most problems happen in the BET and registration steps. My first effort was to apply BET to B0 image of DTI but the binary mask was terrible!
My next idea was to use FLIRT in order to register my DTI into its anatomical then register the previous step result to MNI space and finally use the MNI brain mask template instead of using BET output.
It sounds reasonable but this approach does not work for all my subjects. In some subjects (which are not few) I see a really bad and recurved image in the FLIRT result and in some subject this method work very well. I am new in DTI and I have read several papers but I still can't get what the best pipeline is for DTI preprocessing and how practically the image registration / brain extraction step is correctly performed.
I do appreciate if you can share with me your thoughts,
Bes,
Karen
|