Setting up longitudinal TBSS analysis using SwE for FSL

[yann...@gmail.com]

Dear Bryan and Tom,

I used the SwE toolbox for SPM with success in the past, but struggle a bit with the FSL version. I followed advices provided few years ago on the group (to Alison back in Jan 2021), but still unlucky and I was hoping someone on this group could help me troubleshoot my problem(s).

A bit of background. My study has 3 small groups (16, 14, and 9; total N=39) with DTI scans at 2 timepoints (~6 months apart, no missing data). I ultimately want to determine the main effects of group, the main effect of time, and their interaction on whole-brain FA maps, as well as potential post-hoc tests depending on the results from these effects.

I created the .mat and .con files using the GUI; the .sub file was created manually (see attached). Note that all these files have been stored in a separate RM_ANOVA folder within the stats folder.

However, when trying to run the analysis (using TFCE), I received the following error:

swe -i all_FA_skeletonised -o swe -d RM_ANOVA/RM_ANOVA.mat -t RM_ANOVA/RM_ANOVA.con -s RM_ANOVA/RM_ANOVA.sub -m mean_FA_skeleton_mask --wb -n 999 -T --corrp --modified
swe options: -i all_FA_skeletonised -o swe -d RM_ANOVA/RM_ANOVA.mat -t RM_ANOVA/RM_ANOVA.con -s RM_ANOVA/RM_ANOVA.sub -m mean_FA_skeleton_mask --wb -n 999 -T --corrp --modified
Setting up the model: ERROR: Program failed, unknown exception

Exiting

My best guess is that the .sub file is not configured correctly, but I am totally unsure this is the case.

Does anyone have an idea of what I did wrong here?

Many thanks in advance.

Best regards,

Yann

[Thomas Nichols]

Copying reply from FSLlist...

Dear Yann,

Sorry for the delay.  I've examined the files you sent and have identified the problem:  There is an inconsistency between the contents and the headers of the wav file format.  Specifically, while both .mat and .sub have 78 rows of data for your 78 scans, the header of RM_ANOVA.sub has:

/NumWaves       2
/NumPoints      39
/Matrix

which is the wrong number of columns ("Waves" - should be 3) and rows ("Points", should be 78).

Please review each of your files and make sure that the data and headers are consistent. I assume the (admittedly cryptic) error is because it was expecting 78 rows of subject data and 3 columns of indicators but only found 39.

Otherwise, as far as I can tell, the *content* of your files is fine. Let me know h

-Tom

[Yann Quidé]

Thanks Tom.

As responded to the FSLlist, it ran just fine.

Sorry for the cross-posting, I thought the group discussion was not monitored anymore.

Many thanks for your help!

Yann

--
You received this message because you are subscribed to the Google Groups "SwE-Toolbox Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to swe-toolbox...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/swe-toolbox/CAJoTcz7sFKNH_B4Hnj1vzvONJ3ty%3Dm6k3ETwmH41jQEUb%2BaWHg%40mail.gmail.com.