Image Data to Point Cloud

Hi everyone,

I’m searching for a way to create a semi-random point cloud thats density in specific areas is defined by an underlying CT image. E.g. in areas with an high HU-value the point cloud should be dense as in low HU-value areas it should be light.

Is there maybe already a filter for that?

Thanks, Sebastian

You can iterate through the voxels of your CT image and add/not add a point at that voxel position with a randomly, with probability of adding a point depending on the voxel value.

What is the clinical application?

Thanks for the answer!
After some reading, my current approach will be to use something like the
Floyd–Steinberg dithering algorithm.

As the clinical application, I’m trying to create bone models with a realistic spongious bone structure. Therefore, I’ll probably use the algorithm above, then apply the Delaunay 3D filter for connecting those points and build a grid from those connections, which strength/properties hopefully somehow will correlate with the real bone.

But I’m open for better approach ideas!

Microstructure of the bone is typically several magnitudes smaller than size of the mesh elements that you would want to use for FEA. To achieve different mechanical properties in different mesh regions you can vary material properties across mesh elements.