Hi Stephen-
I mostly agree with the previous responses- BET can produce reasonable rhesus skull strips but you need to tweak the settings. It is then about 95% accurate for most subjects, with a few spectacular failures. For some applications, especially PET, 95% is not good enough. Having said that, BET is still probably the best automated tool for this application.
One thing to add is that it is fairly sensitive to image quality, especially tissue inhomogeneties, such as a bright spot near the center. This makes for different levels of satisfaction from one scanner to another. If you have such artifacts, you will need to perform some sort of correction... which usually entails first getting a good brain mask. Since the images need manual touching up anyway, I cut to the chase by hiring a few undergraduate students to manually draw whole-brain ROIs. This gets me up to 99+% accurate, with very good consistency.
Cheers,
Terry
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Terrence R. Oakes
Waisman Laboratory for Brain Imaging and Behavior
University of Wisconsin-Madison
[log in to unmask]
http://brainimaging.waisman.wisc.edu/~oakes/
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
----- Original Message -----
From: "Stephen J. Fromm" <[log in to unmask]>
Date: Thursday, October 2, 2008 9:21 am
Subject: [FSL] BET in rhesus monkey
To: [log in to unmask]
> I'd like to use BET to skull strip some FSPGR images of rhesus brains.
>
> Can I just apply the tool naively, or are there special considerations
> I need to
> make to account for these not being human?
|