registration fails, is scale the problem?

[Stein]

Dear Scalsiomo comunity

I am working on a project to create a SSM of the radius (the bone, not the circle property).

Most of the code is done, however, the correspondence does not work yet. The problem is that, especially if the radii differ in length, the shape of a reference bone morphs to a local minimum as can be seen in the image (I use paraview for visualisation). All the point of the reference distal radius morph to the shaft of the radius. 

Although I wanted to remain the scaIe in my model, tried to solve this problem by using similarity3DLandmarkRegistration(and eventually scale back when correspondence is found.) However, this similarity transform does not work properly as can be seen in this image. I think the good matching points on the shaft prevent the shape from scaling down

Does anyone have any advice to solve either the registration error or to get the similarity transform going?

Kind regards,

Stein van der Heide

[Stein]

Oh and adjusting the scale of the reference bone instead of the target bone as done above, resulted in the same problem.

Op zaterdag 8 januari 2022 om 23:51:35 UTC+1 schreef Stein:

[S Arbabi]

Hi Stein,

To eliminate the effect of bone length, you can:

- if you're concerned mostly about the subchondral parts, crop the bones for instance the way it's done in the study by Lynch et al 2019.

Lynch, J. T., Schneider, M. T. Y., Perriman, D. M., Scarvell, J. M., Pickering, M. R., Asikuzzaman, M., Galvin, C. R., Besier, T. F., & Smith, P. N. (2019). Statistical shape modelling reveals large and distinct subchondral bony differences in osteoarthritic knees. Journal of Biomechanics, 93, 177–184. https://doi.org/10.1016/j.jbiomech.2019.07.003

-or you can try to limit the area of alignment to some landmark more in the subchondral area, for instance you can do that manually putting a few landmarks on the subchondral area of all meshes and do a landmark alignment, or you can just try to make sure all the landmark correspondence that you're making in the ICP alignment phase are not somewhere on the length of the bone.

Best Wishes,
saeed

[Stein]

Hi Saeed, 

Thanks for your answer. It is truly amazing how you guys maintain this community and help us with our problems.

I hoped to do it without manual landmarks, but when I used it, it worked great! Thank you for the help!\

Best regards

Stein

Op zondag 9 januari 2022 om 13:31:34 UTC+1 schreef gsar...@gmail.com:

[S Arbabi]

Dear Stein,

Happy to hear that.
Have a good time.

Best wishes,
Saeed