Thanks Saad for the quick answer.
I fully get your point.
The attempt to [calculate the probability that A connects to B via C vs. the probability that A connects to B without passing through C] is an option; however, here one would also unwantedly retain samples that circumvent a structure of interest, rather than focussing only on samples "going through" it -> consider a tumor/its vicinity in the pyramidal tract: some tracts are pathologically discontinued, others, however, "make" it down stream, while even others invervating e.g. the exact same muscle take a route slightly outside the tumor mask.
Then I figure the only theoretical way (though perhaps not being very practical, with e.g. large irregular masks) of looking at the proportion of samples "terminating" in a given structure vs. continuing (i.e. comparing histograms) would be to compare samples given the condition "terminate at a mask n" vs. "terminate at a mask n+1".
Regards,
Stefan Kreisel
|