Hi,
I have trained my femur geometry using 512 particles with shapeworks software. However the raw and reconstructed geometry shows the deviation in the geometry.
Can you describe a bit more about the deviation? I assume you are finding the reconstructed surface and groomed surface differ? Is it a consistent deviation? Or a deviation in only some particular areas?
We have some visualization tools that can assist in examining this:
Thank you for your reply.
Yes there is deviation in groomed surface and reconstructed surface. I can see deviation is there even if we increase the number of particles. It is not coming to zero. I am attaching one image of the deviation analysis. I hope now you can answer the questions below,
It may take a very large number of particles to reduce the deviation. But there are other factors. The reference template mesh geometry is taken from the median shape and warped to the per-sample particles. There are several input-dependent factors that will each have an effect here. There’s no way for us to say how much deviation is expected as it is input dependent.
