On 14 Jul 2009, at 03:32, Long Chuan wrote:
Hi, FSL experts
I am trying to map tracts of the corticospinal tract (CST) from multiple subjects to a common space. By reading the previous threads about this topic, I did the following: I first transformed the CST traced from each subject to a template and then normalized the CST tracts using waytotal. Then, the CSTs across subjects were added together, and lastly, they were averaged to form a CST map from the population. My questions are as follows:
1) What is the best way to interpret this map? From the previous mails, after the normalization by waytoal, it seems that the tracts can be interpreted as the tract probabilities given the prior knowlege that the tract is there. Can I interpret the map from multiple subjects as
the CST tract probability at a population level?
Yes I think that's right.
You can convince yourself that it is true by writing:
p(A->B) = p(A->B | Y1)p(Y1) + p(A->B | Y2)p(Y2) + p(A->B | Y3)p(Y3) + ...
Where Y1, Y2, ... are data from different subjects, which are equally probable a priori: p(Y1)=p(Y2)=...
2) One of my interests is to look at if the CST in the two hemispheres have same 'tractibility'. After running tractography on each individual, I did find that it seems left CST is more tractable than the right in my population, which usually has larger waytotal than that on the right side. But if I normalize the two CST using waytotal (left waytotal is larger than right waytotal), this asymmetry will be gone in the pouplation CST map. So, my questions are i) if this asymmetry in 'tractibility' in the two hemipsheres noticed in the FSL tractography worth reporting; ii) if it does and it will not show in the map obtained using above-mentioned method, what is the best way to characterize this difference? Using 2 sample-ttest on the
waytotal or some other methods?
When you say that you have observed asymmetries in the CST, do you mean in terms of the spatial extent of the tracts? Or do you mean in terms of the voxel-wise probability values along the tract?
I there is no difference in the tract extent, then normalising by waytotal should keep an asymmetry in the pdf, whereas when there is a difference in the spatial extent, the asymmetry after normalising will be lower.
The easiest asymmetry to report is probably an asymmetry in the spatial extent of a tract that is defined using waypoint masks that are equivalent between hemispheres.
Whether it is interesting or not I guess depends on what it means. It surely opens up other questions, such
as:
- is the asymmetry specific to the part of the CST that you have dissected? [presumably the hand M1?]
- is it a mere asymmetry in the shape, or are there micro-structural asymmetries as well (FA,MD,etc..)?
- does it have functional/behavioural consequences/correlates?
- more precisely: does the degree of asymmetry, as measured using this method, predict the asymmetry in the subject's motor functions?
- etc. I am sure there are plenty of interesting (?) questions here...
Cheers,
Saad.
Sorry for some mnay dumb questions, and I would appreciate your help :)
Longchuan
