|
From: Andy A. <ad...@sc...> - 2012-02-15 17:18:12
|
Hi james, Thanks for this work. Are these licensed under the GPL so that we can include them in EIDORS? Where would you recommend they be placed? I see where the rectangular tunnel code would go. Could you give an example of how to use the gmsh code? For the compressed sensing, I've seen a few people think about using it for EIT (and other soft field tomography modalities). To be perfectly honest, I've never been convinced that EIT has the sparsity required to make compressed sensing a good idea. However, I would be interested in seeing what you get ... -- Andy Adler <ad...@sc...> +1-613-520-2600x8785 On 9 February 2012 17:56, James Snyder <jbs...@fa...> wrote: > I've made some improvements to the GMSH mesh reading code which allow > for reading in 3-D meshes (among other things). I'm sure there are > some bases that might not be covered in it, but it does parse the > ASCII .msh files in a manner more closely to how the Gmsh 2.5 manual > indicates, and passes through surface numbers so that one can select > surfaces based on them. There are a few variables that it will not > fill in like: > edg = []; % not sure what to do with this, I don't have a format > description for it > bc = []; % this looks like it attaches numbers to boundaries of the > mesh, I'm not 100% sure how I should be treating this differently than > fc > mat_ind = []; % It looks like this is sortof the equivalent to fc > (face numbers) but for volumes? I should be able to use physical > groupings for this > > https://gist.github.com/1764337 > > gmsh_mk_fwd_model still needs some modification in order to be more > usable, for example, in selecting electrodes automatically or > manually. It would make sense to maybe have physical groupings that > it could use for cues in this. I can update that to the gist if I get > around to it. > > As mentioned before, I could also provide an example of doing manual > 2-D meshing with NetGen. I've gotten that to work with an adjusted > geometry and I think I have a set of assumptions that may work fairly > generally: > https://gist.github.com/1784027 > > One additional point is that I've started playing around with some of > the compressed sensing algorithms described in this paper: > http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5627165&tag=1 > > The authors provide the parameters they used with a couple of > different algorithms, and I've been able to add *_inv_solve functions > for a number of them, and have gotten them to work with some success, > though they're dependent on various solver code packages that > different research groups have provided for MATLAB. Does anyone on the > list have experience with these? > > -- > James Snyder > Biomedical Engineering > Northwestern University > http://fanplastic.org/key.txt > ph: (847) 448-0386 |