pyeit-1.1.9: Generate realistic head and lung meshes!

[liubenyuan]

What's Changed

• Improved f solver calculation by @ChabaneAmaury in https://github.com/liubenyuan/pyEIT/pull/37
• Optimize .et3 and .mes loading routine. Generate optimized thorax shape using distmesh and shapely by @liubenyuan in https://github.com/liubenyuan/pyEIT/pull/38
• Generate High quality Head (symmetric) and Lung meshes.

Full Changelog: liubenyuan/pyEIT@v1.1.8...v1.1.9

---

This discussion was created from the release pyeit-1.1.9: Generate realistic head and lung meshes!.

[DavidMetzIMT]

Hy,
Thanks for the new realease! About the improved vectorized calculation of the f solver, I assume this was done to improve the calculation time of the static reconstruction of jac gn! Would it make sense to also add this functionality to EitBase as an attribute to improve the calculation time during the new instantiation of jac, bp and greit solver (especially for bigger mesh)?

[liubenyuan]

The vectorized PR was made by @ChabaneAmaury for accelerating forward computing. It is not implemented on image code yet. But sure, it is very valuable to do so and I will create a Issue on it.

I tried to test forward solver using NY-HEAD 3d model (a very big, refined head mesh) but still failed to load the complete model into RAM. I will working on it for the next releases.

[ChabaneAmaury]

Hi,
Would you give me some insights on what exactly you would like to improve?
I will try to take a look at it in the next days.

[liubenyuan]

I try to load the NY-HEAD model: https://www.parralab.org/nyhead/ (where the .inp file is used) and placed electrode on the boundary, do the eit forward simulation.

As to the inverse imaging, I think a 3D inverse imaging is not feasible. The Jacobian should be calculated on a 2D and the unkowns are suffice for the inverse problem.

[DavidMetzIMT]

Hy,

Would you give me some insight on what exactly you would like to improve? >> was it for me?

Basically, you add your vector as an *arg in the EitBase constructor, save it and use it when instanciating the solvers jac, bp and greit. Nothing really special. These call your improved solve_eit method. I've already worked out this implementation, I will make a PR after I have tested it!