Back to the main page.

Bug 3014 - ft_prepare_mesh: functionality for calling isosurface on non-boolean volumes

Status NEW
Reported 2015-11-25 12:45:00 +0100
Modified 2015-11-26 09:34:18 +0100
Product: FieldTrip
Component: core
Version: unspecified
Hardware: PC
Operating System: Windows
Importance: P5 normal
Assigned to:
Depends on:
See also:

Jan-Mathijs Schoffelen - 2015-11-25 12:45:16 +0100

This is the larger perspective on the issue encountered with Tom and Lau, where the 'isosurface' method (or more generally prepare_mesh_segmentation) by construction only works on boolean volumes, which leads to 'granular' meshes. The quick and dirty proposed fix (allowing for a last step smoothing of the boolean image prior to calling isosurface), has been discarded as a clean functionality. Rather, we should investigate the possibility to cleanly pass a 'probabilistic' segmentation (in Tom's and Lau's case, a probabilistic version of the scalp) to ft_prepare_mesh, and check whether the different methods are capable of working with this. If this would work, we keep the responsibility for creating a proper segmentation with the user (where admittedly FT's scalp-segmentation always yielding a boolean volume is suboptimal).

Robert Oostenveld - 2015-11-26 09:34:18 +0100

Right now the segmentation and meshing is mainly developed with the application of BEM volume conduction modelling in mind. Keeping the segmentation and meshing implementation clean and making incremental improvements is also relevant for future improvements in source modelling, such as using FEM (where the mesh should also implement the fact that the head is not a single closed volume, but it contains multiple air-filled compartments).