[FieldTrip] ft_volumerealign: issue when coregistering to headshape

[jan-mathijs schoffelen]

Dear all,

I have fixed a somewhat critical issue in ft_volumerealign. Please disregard this e-mail (i.e. don't worry about it) if you have never used this function to coregister your anatomical MRI to a headshape (with cfg.headshape = something) in the past couple of months (starting from October 2013).
The long story short: when supplying ft_volumerealign with a headshape in the configuration, the function tries to register the anatomical MRI to this headshape, by creating a 3D model of the scalp surface (based on the MRI) and using an iterative closest point algorithm for registration. So far so good. Yet, as of revision 8576 (committed to svn on Sept 30 2013) I added an interactive alignment step to this procedure, because the icp-algorithm is known to behave well only if the point clouds are already approximately registered. That is, in my experience an approximate registration based on the fiducials only was not always sufficient to achieve a nice coregistration. This being said, the introduction of this additional interactive step also introduced a bug, in that the transformation matrix that was estimated with the icp-algorithm was not properly dealt with. Actually, this information was never used for the registration and as a result the coregistration matrix outputted in ft_volumerealign was the one that resulted from the interactive realignment only. Not a total disaster, because I expect the user to get an as good as possible coregistration by hand to begin with, but also not how it should be. My apologies for any inconvenience caused.
The issue has been fixed as of svn revision 9096.

Happy computing,
Jan-Mathijs


Jan-Mathijs Schoffelen, MD PhD 

Donders Institute for Brain, Cognition and Behaviour, 
Centre for Cognitive Neuroimaging,
Radboud University Nijmegen, The Netherlands

Max Planck Institute for Psycholinguistics,
Nijmegen, The Netherlands

J.Schoffelen at donders.ru.nl
Telephone: +31-24-3614793

http://www.hettaligebrein.nl

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20140114/c8752b14/attachment-0001.html>