[Femur Miranda 2010] Determination of point P2 is not as described in the original paper about msk-staple HOT 6 CLOSED

The original paper does not mention the potential orientation error of the initial inertia axis, so, in order to find which parts of the femur is anterior or posterior, we look to the closest point to the bounding rectangle on the cross section, assuming this point is posterior. When you look for point P2 we know that it must lie on this side of the cross section. @modenaxe do you want me to comment the whole function, or are you already working on a alternative version ?

from msk-staple.

I realised the non-directionality of the inertial axes was an issue, but could not properly understand the rest of the implementation. Maybe have a look at the function I was working on at this link. If you can add some comments would really help me. You can easily find the section because I marked it to rewrite it. If I understand it then I am happy to keep it! Thanks @renaultJB !

from msk-staple.

by the was @renaultJB , the Miranda Femur function does work properly - so it was correct.

from msk-staple.

I have added some comments and made minor code change to clarify the method. One thing that differ with the method of the original paper is proper polygon or polyhedra to axis intersection is not implemented. As in my test dataset the resolution of the mesh was quite high (~0.5mm) I didn't implemented intersection point identification functions but just identification of the closest point on the polyhedra (mesh/triangulation) to a given axis. I think this method gives good results as long as the mesh is sufficiently fine and present triangles of similar size.

from msk-staple.

Awesome! Maybe write a note on the function reporting this same comment, so if someone else uses it he/she will know that we are aware of the issue

from msk-staple.

Using Miranda's code received from the Authors.

from msk-staple.