On Mon, 13, Sep, 2010 at 05:53:51PM +0100, Chris Racey spoke thus..

Actually, after my last message, Padraig came to see me and mentioned another option -R which iteratively repeats the BET procedure from different starting points. He suggested that this may achieve the same improved result. I tried this option instead and indeed it does still seem to be an improvement. However it runs considerably quicker and doesn't seem to result in intensity changes in the image.

That was going to be my suggestion too. I don't now normally run bet without using the -R flag. As Andre says, changing the starting co-ords can also help - for the record, the standard automatic cleanup for the structural scans currently uses -c 90 116 166 but doesn't use -R. I'm not entirely sure why it doesn't as I thought it did (and was about to say it did until I double checked).

I'll probably change this to use -R by default (for the automatic ones) unless someone can remind me why it doesn't...

Mark