Hello All,
I have been playing around with FAST for a while now trying to get good
results using Brainweb but to no avail. I was hoping someone would be so
kind as to point me in the right direction as far as improving my results.
So far I have tried FAST using the output of BET and 3dSkullStrip. Using
various combinations of the parameters (apriori maps, partial volume
extraction, etc.) I have been able to achieve a misclassification rate (MCR)
of approximately 50% for CSF, 37% for GM, and 32% for WM. I have seen other
papers where the MCR is very very low (close to 4%). Would someone please
shed some light on what I am doing wrong, please? Thanks in advance!
Best Regards,
On Tsang
MCR is defined as 1 - (volume(A&B) / volume(B)) where A is the segmented
output of the algorithm that is being tested and B is the ground truth
provided by Brainweb.
|