Text archives Help

---

Hi Scott,

If you are looking at a CT scan, then yes, the X axis is the radiodensity. The "Y" axis is gradient magnitude, but lets just call it "another variable" for now. Then I can explain that the gray means how often that combination appears in the data set. That is, if that particular (radiodensity, GM) pair occurs very, very often, than you see a white pixel for that location. If it doesn't occur at all, you see a black pixel.

This can give you a quick idea of how changing a transfer function will affect your visualization. Creating polygons in the black regions should have almost no effect; conversely, creating them in white regions will make large amounts of the data visible.

Back to the Y axis: it is the gradient magnitude. More precisely, we calculate the difference between each voxel and its neighbors. The difference between the current voxel and the voxel to the left defines the "pull" to the left; likewise with the current voxel, the voxel to the right, and the "pull" to the right. All of these are added together, such that we get a single direction that the voxel is 'moving'. We already have a radiodensity for that voxel, and now we have the gradient, and so we have a (radiodensity, gradient) pair, which is used to perform the lookup into the above structure.

The gradient magnitude is not strict unary function of the radiodensity for two reasons: error inherent in scanning, and the location, and therefore the neighboring materials, differ. The first is easy to explain: a scanner isn't perfect, and so the response for "muscle" (e.g.) is not always going to be (say) "40". Sometimes it'll be 39. Or 42. etc.

As for the location, consider a volume of tissue in, say, your hand. Some of that tissue is going to be riiiiight next to the bones. Some of that will abut blood vessels. Even ignoring the noise issue, the values for the tissue, the vessels, and the bone are all unique. Therefore the change in values at the interface between tissue+vessels will be unique from the change in values at the interface between tissue+bone. To be concrete, if tissue is 5, vessels are 10, and bone is 40, then the tissue/vessel interface will have a gradient magnitude of 5, whereas the tissue/bone interface will have a gradient magnitude of 35.

The pair for the first is then (5, 5), and for the second it would be (5, 35). Therefore lower in the 2D transfer function you could highlight the tissue/vessel interface, and higher you could highlight the tissue/bone interface.

Does that help?

-tom

On 21.09.12 23:14, Scott Krehbiel wrote:

Hi,
I'm looking at the example scenes, and I'm a bit confused.

I understand that the X axis seems to be the radiodensity of the sampled
location.  In the CT scans, Bone is further to the right, and skin is
further to the left.  I don't yet understand the Y axis though. Wouldn't
everything at the same X coordinate have the same density?

If radiodensity is mapped to X, then what is mapped to Y?  Is it the
local average density - ie; how dense are the surrounding voxels?

Any help is much appreciated
thanks
Scott

---

• [IV3D-USERS] Y axis of 2D transfer function editor, Scott Krehbiel, 09/21/2012
  • [IV3D-USERS] Re: Y axis of 2D transfer function editor, tom fogal, 09/21/2012

---

Archive powered by MHonArc 2.6.16.