BET

Overview

BET takes an image of a head and removes all non-brain parts of the image - it is a useful tool to use as a preprocessing step before further anlaysis. BET uses a surface model approach, to robustly and accurately carry out the segmentation in less than 1 minute. It normally requires no user intervention. BET has been tested on thousands of images, giving accurate segmentation in all but a very few cases.

BET2

A new version of BET has been developed comprising two programs:

• bet2 - a direct recoding of bet into c++, with close to identical results
• betsurf - using a registered T1 and T2 image pair estimates inner and outer skull surfaces and outer scalp surface

Papers

A paper on BET has been published in Human Brain Mapping. For detail, see a report on BET (PDF) or a slideshow.
An abstract on BET2 was presented at HBM'05. For more detail on BET2 see the BET2 technical report.

Software

BET and BET2 are available as part of the FMRIB Software Library.