search

mne-tools / mne-python

MNE: Magnetoencephalography (MEG) and Electroencephalography (EEG) in Python
https://mne.tools
BSD 3-Clause "New" or "Revised" License
2.7k stars 1.31k forks source link

ENH: Anatomical template warping #3527

Closed larsoner closed 8 years ago

larsoner commented 8 years ago

I'm starting to read a bit about anatomical template warping, and it looks like there are some standard methods like TPS warping or warping based on spherical harmonics that can be used. Any thoughts on the best ones to use?

@ftadel you use the TPS warping technique in BrainStorm, right?

ftadel commented 8 years ago

Hi Eric,

We use code written by John Mosher and John Ermer: we start by computing a smooth surface based on the digitized head points (using spherical harmonics), then we compute a deformation field based on the two surfaces (the one from the MRI and the one from the Polhemus) and apply it on the surfaces and the MRI. It is done in function brainstorm3/toolbox/anatomy/bst_warp.m (yes, I'll move my code base to github soon :)

The corresponding tutorial: http://neuroimage.usc.edu/brainstorm/Tutorials/TutWarping

Francois

larsoner commented 8 years ago

Thanks @ftadel. @christianbrodbeck would this be useful for times when you want to use a surrogate? Any thoughts?

christianbrodbeck commented 8 years ago

Hm yes warping might potentially improve the localization for surrogates if the head shape is accurate. Although the advantage over scaling would depend on the head shape reflecting brain shape. Do we have an idea how well brain shape is correlated with head shape?

larsoner commented 8 years ago

IIRC both of the papers linked above do tests related to that and show increases in inverse accuracy