Dear FSL experts,
I posted this topic several days ago and Stephen gave me some nice suggestions and help. I am so grateful to him.After these days' thinking I have some more questions about this issue . I understand you might be very busy in the FSL course and the conference (SfN) recently. Hope someone can answer me.
I raised this issue because my reviewer doubted the difference in the overall morphology of the cranium of two groups of subjects in my study might influence the results (TBSS) and asked me to contrast the degree to which the TBSS software changes original tract locations between the two groups. After Stephen's help and reading their study in 2006, I understood TBSS uses two steps of registration to exclude the "system difference in extract tract location between groups of subjects biases the comparison of FA values between the groups". It might not be necessary to calculate the degree of registration and contrast it across groups. However, for more perfectly answering reviewer's question and i myself also am curious whether the degree of registration in two groups of subjects (from China and U.S.) is different, I am trying to do it.
I met some specific questions when I calculate the degree of registration:
1. How to get the index of registration in first steps? I saw 3 volumes represented x y z displacement (right?) from TBSS analysis. I thought there should be one associated parameter, right?
2.Someone told me to use Jacobian determinants, is that ok? and use the script "fnirtfileutils" to convert the coefficient-file into Jacobian determinants?
3.I also thought about using the square root
of x2+y2+z2 to index the registration, is that ok? And is there a script to convert the coefficient-file into that value? I used fslsplit to split the 4D warping file into x y z file and then use fslmaths to calculate them. Finally, for every subject, there is one 3D file. But it's very strange I can not merge all files (from each subject) into one 4D file for further analysis (I want to use fslmeants to extract the registration value of each tract for further contrast). They are different in matrix!
4. The most difficult thing is to get the "warping" of the second step. Stephen told me to use tbss_skeleton -d and look at the source code to get more information. The source code is hard for me. Can someone give me some specific instruction about how to get the warping of "each subject projecting into the mean skeleton"?
I might think about the registration too much, but I really want to make clear of this. Please let me know if you think I am not correct about what I mentioned above. Very hope someone would answer me.
Thanks for you guys' always help!
Mingxia Zhang
