Dear all,
I am trying to investigate the effect of an additional preprocessing step on the results of the TBSS analysis. Specifically, I am most interested in whether it will affect the two groups of participants (patients, controls) differently.
In order to choose a proper permutation scheme, I was previously recommended by members of this list to check for this interaction-type effect by performing a separate analysis for difference images (usual image - additionally preprocessed image) and compare those between the two groups with a simple t-test. I subtracted the two images for each participant registered to the common target image in MNI space (determined in previous TBSS steps).
However, TBSS does not produce sensible results using these subtracted images. I am suspecting it could be related to the range of values in the resulting maps (parts are negative and they are much lower than usually). If I add the target image FA map (selected using tbss_2_reg -n) in MNI space to all the difference images, the results are sensible and in accordance with what is expected.
I am wondering whether this analysis should be performed differently. Do you perhaps think that in stead of adding the target image I should rather add a mean FA map of all images (generated with fslmaths) or perhaps perform an alternative mathematical transformation?
Thank you for your help and all the best,
Rok
|