I've had to do this kind of thing in the past, trying to use an STL file to create a model. I didn't use the STL directly for anything. An STL is just a huge set of planar facets. Any "curve" obtained from it is not a mathematical curve, it's a bunch of tiny line segments.
What I ended up doing is modeling the object in question directly, using measurements from the STL. Using sets of points to find radii, lengths, etc. I assemble my nice geometrically "clean" model with the STL model and check that I'm getting things right.
The model you are attempting to build seems like more or less a simple revolved solid, except for the spiral(?) passage. If it was me I'd probably take the approach of building the main outer body, then attack that spiral passage maybe with equation driven curves. That way you'll be able to get the geometry to be more rigorously defined, and it will allow you to, if you wish, make variations, make a larger version, etc.
