Dear TBSS experts,
I have a question concerning the process one should follow during a TBSS analysis. It is all very straight forward in the online manual but however there is one sentence, in the tbss_3_postreg part that is troubling me.
It says : ''If the registration has worked well you should see that in general each subject's major tracts are reasonably well aligned to the relevant parts of the skeleton. ''
I have previously posted a question regarding how strict one should be when examining the amount of warping of the non-linearly registered FAs in comparison to the mean_FA_skeleton. Stephen Smith kindly replied that their is indeed a certain level of subjectivity involved in this process.
What I would like to ask now, directed mostly to those who analyze pediatric and neonatal data sets, is what is the process they follow when they examine the all_FA folder?
There are numerous articles with both pediatric and neonatal subjects out there but none, at least none that I have read, describes the number of subjects they start of with in their analysis and the ones that were possibly excluded due to warped FA maps in the all_FA folder.
So basically what I am asking is: Do you fellow researchers look into the FA images of the all_FA folder? If you see mis-registered FA maps which have been warped, due to the registration and do not align to the skeleton, do you exclude them when you set up your Glm? Or do just leave them in the analysis and present the results you find?
Additionally I am curious about the impact such warped FAs have on the subsequent statistical analysis. Is leaving them in the analysis or excluding them from further consideration a ''statistically'' significant factor that can lead your results?
I would be grateful if you could possibly share your thoughts on the above questions.
Thank you beforehand for your efforts and insight so far,
Georgios Alexandrou MD
Karolinska Institutet,
Sweden
|