Hi Steve,
I had a few question regarding the symmetry script you had written, and the
methods used to derive the asymmetry values.
My main question relates to the step where the dilated original asymmetrical
skeleton is used to mask the original symmetrical skeleton. In selecting
data where structures are already more or less symmetrical, is one not
losing information that one is specifically interested in, because we know
that the brain is not perfectly symmetrical.
Related to this - is the following method valid: Using a symmetrical
template, normalising flipped and unflipped images to this, deriving a
symmetrical mean_FA image & skeleton and then finally deprojecting
flipped/unflipped images onto this skeleton ? I know you have commented on
this before but I am still a little unclear as to the shortcomings of this
method.
Thanks.
Mahinda
|