Skip to content

Latest commit

 

History

History
39 lines (31 loc) · 2.05 KB

README.md

File metadata and controls

39 lines (31 loc) · 2.05 KB
Raw

Headmodel Individualization

Data-driven head model individualization from exact electrode positions or photogrammetry of the subject's head surface improves M/EEG source localization accuracy!

Individualization algorithm

pca_surfacemesh_warping.py contains the main function of how to start the inidividualization. It is based on a low-dimensional representation (PCA) of head shape surface meshes trained on a equally segmented and triangulated MRI database of 316 subjects. Warping is done by finding weights for the PCs by minimizing the shape difference between electrodes / scalp proxies and fitted scalp. It contains the following steps:

  • Upload your fiducials (NSA, LPA, RPA) and scalp proxy, i.e. sensor positions or any points of the scalp. Ideally points above the ears.
  • Using the fiducials, transform the points to the CTF-coordinate system, since the database on which the PCA was applied, lives in CTF space.
  • PCA warping, parameters are: number of PCs used for reconstruction, regularizer type (if meshes are intersecting)
  • Transform back from CTF in original input space.
  • Save surface meshes as .tri (for OpenMEEG) and as .mat for Matlab/FieldTrip or any other mesh format like .stl, .obj, .ply, ....