ITK was lacking IO filters for meshes formats. This publication implements a mesh IO mechanism, mirroring the Image IO mechanism already in ITK (factory pattern), as well as factories for several formats (OFF, freesurfer, VTK polydata binary, ...)

The contribution includes several files with different formats. The tests read and write the files, using either the itk::Mesh or itk:QuadEdgeMesh data structures, and compute the errors between the input and  the output files. The error is below the threshold (1e-06) and all test pass on my machine.

Source code, and data are provided along with tests to reproduce all the results announced.

I downloaded, compiled and tested the code. Everything went fine except for the small known glitches due to my using itk v4, latest revision, which is hardly due to the author. I think, even though I did not test, that the code should run fine with 3.20.

The paper could be more explicit on pixeltype (data type ), their implementation of the polylineCell and the implicit point numbering scheme. Nothing that would prevent an user from using the code, but as a developper, I had to double guess.

ITK quality level. Some typos or left over from here and there (polylinecell still has commented out code, and reference to trianglecell or polygoncell for example.) Overall very good.

Did not check.

This will be usefull to everybody working with meshes, and there are a few of those in the NAMIC / ITK community. It is a feature that had been missing for too long.