The process is pretty straightforward:
I have all the CT-images in a folder (let's say 100). Then with copy-and-transform I make 100 polygon-planes (highly subdivided) with equal distance on top of each other.
Using a for-each-loop I just iterate through every plane.
With attribfrommap I map the 1st image to the 1st plane, the 2nd image to the second plane etc...
So now all planes have a density attribute.
Then connect the stack of planes to a volumerasterize-attributes and sample the density. This then turns your poly-/point-geo to voxels.
Now you can apply all the volume-goodness to it or just convert to a poly-surface.
Hope that helps!