I have to stress that when constructing a transformation using the vertices of STL surfaces (AnyFunTransform3DSTL), the surfaces have to be topologically equivalent, i.e. the surfaces have the same number of triangles and each neighbour and vertices represent the same features on both surfaces.
So if you do not have this your best bet will be to continue with the AnyFunTransform3DRBF function.
Looking at your code with the RBFDef, it looks fine. You can try using the settings I pasted below and see if there is an improvement. Or also try to select more points to see if that makes a difference.
AnyFunTransform3DRBF RBFTransform = {
PreTransforms = {&.AffineTransform};
RBFDef = {
Type = RBF_Triharmonic;
Param = 0.2;
[..rest of code..]
};
You should also bring in your target bone and draw it in another color to see how well the morphing is going. Make a little chunk of code (note: you might have to adjust file names/locations and ScaleXYZ accordingly) like this within the HumanModel.any file:
I’ll try to select more points to see what will happen?
As you mentioned above about the AnyFunTransform3DSTL, I am wondering how can I do to acquire the topologically equivalent surfaces? Are there some tutorials or software or something else I can reference?
I am sorry for the late reply. There are some tools to do it. I believe Mimics by Materialise may have a morphing tool. Another option could be to generate bones using a template mesh from statistical shape models.
After custom scaling the femur and tibia based on CT imsges, I found the anatomical reference frame and the kneejoint reference frame are not even perpendicular to each other (X,Y,Z). Should I define the new reference frame?
The blue bone is the femur exported in local fame from AnyBody and the green one is the adjusted implant which shares the same coordinate system with the femur.
The bone in Mimics does not correspond the one in AnyBody. You either exported scaled surface, or exported in global, or using a bone from a different dataset, i.e. bone from TLEM2.1, when the Mimics one is TLEM1.1. If you have everything consistent - you should get right match.
The blue one is the femur exported from AnyBody after custom scaling and I import it into the mimics 3-matic software to adjust the implant to match it.
I believe I used the right dataset and exported in the local frame. You mean I shouldn’t export scaled surface?
Good to hear things worked out. There are many different options for surface fitting tools, it depends what you have available and what is easiest for you (3-matic, matlab, python). You can fit surfaces directly in AnyBody by importing your surface selection as an STL and using the GetVertices() function in AnyDrawSurf. In the example below I fit a cylinder to the tibiofemoral contact area of the medial condyle of the femur:
You can do a similar procedure to fit the lateral condyle, and then connect the center nodes to define a the tibiofemoral hinge joint axis. Please note this is only one way of defining the knee joint, there are many options in the literature. With respect to the hip joint you can fit a sphere to the femoral head (using AnySurfEllipsoidFit with spherical switch ‘on’. Finally, please see this source for how the ankle joint was fit:
[1] Parra WCH, Chatterjee HJ, Soligo C. Calculating the axes of rotation for the subtalar and talocrural joints using 3D bone reconstructions. J Biomech 2012;45:1103–7. doi:10.1016/j.jbiomech.2012.01.011.
I just noticed that fibula is not in the GeometricalScaling folder, does that mean that we cannot custom scale the fibula based on CT and it can change just based on the tibia?
Hi Lee, fibula and tibia share a scaling function. You need to implement tibial scaling function in such way that it would morph fibula well at the same time.
