Hi Emily - The weighted average is the most reliable in practice, and it is also the proper statistical definition of mean (expected) value. It weighs the value of FA etc at each voxel with the probability of the tract going through that voxel.
The cases that are missing stats may need to be rerun. I'm guessing that those will look like single curves if you view the path.pd.nii.gz volume. You can rerun them by using "set reinit = 1" in the config file, and also specifying in the config file to run only the specific tract and specific subject. You'd have to rerun the -prior and -path steps only.
a.y
On Sun, 3 Aug 2014, Emily Louise Belleau wrote:
Hello Tracula Experts,
I noticed that for a few of my subjects, I am missing all "Center" statistics for FA, MD etc. for certain tracts like the left uncinate fasciculus for example.
I am assuming this means that an optimal pathway was unable to be found? I do have the average and weighted statistics for these subjects. Should I consider excluding these subjects or should I try analyzing them again with different parameters?
Additionally, I noticed that for a couple of my subjects, the "Length Center" (length of the highest probability pathway) is unusually short compared to the rest of the subjects? Is this a problem and do I consider excluding or tweaking the parameters for these subjects?
Also, do individuals usually report in publications just the average statistics or are the weighted average or center statistics used?
Attached are the stats of the problematic subjects. Any advice would be greatly appreciated!
Emily