air@unix.cis.pitt.edu (Alvin I Rosenthal) (02/20/91)
Sorry, but its time for another novice question. I am posting this to the net in the hopes that someone might be able to provide a little guidance. I am currently thinking about starting a project to do 3d reconstructions of CT scans for teaching purposes. There are several commercial machines available for such purposes but these are both expensive and non-extensible. My goal is to make the reconstructions available on either simple workstations or pc's using extended or super VGA. Since the resolution of a CT scan is typically 512 X 512 this seems like an attainable goal. After the 3d bony reconstruction is done, I would then hope to build a facial musculature on top using methods similar to Waters (Waters, KA Muscle Model for Animating 3D Facial Expression. SIGGRAPH 87:17-24) or Parke (Parke, FI Parameterized Models for Facial Animation. CG&A 2(11) Nov 82:61-68). Once this is accomplished manipulation of the facial skeleton could be done with visible changes in the musculature (Move the bones and watch the face change.) To keep the coding to a minimum I would use as much commercially available software as possible. (Autocad is one possibility, with its C interface in version 11 and Renderman like Autoshade.) The first stumbling block I have reached (probably the first of many) is how to convert the CT data to polygonal surfaces suitable for inclusion in a cad or rendering package. Each CT voxel (volume pixel?) has a predefined size and location. Just using these "cubes" would give a very rough appearance. Scanning for the outer edge and extruding this polyline the thickness of the CT cuts is unusable because the interior cannot be represented this way (The interior of the skull is very complex with sharp projections, sinuses, etc.) Other possibilities such as using the vertices of the bone surface voxels as control points for a bicubic patch seem unneccesarily complex. What I would like to do is characterize each bony surface as a smooth set of polygons , both the face of the voxel and the wedges or polygons connecting stepped voxels. If anyone knows of any commercial programs, shareware, public domain programs, or has any suggestions for algorithms I would be very appreciative. Thanks in advance -- Al Rosenthal air@unix.cis.pitt.edu
lewis@cortex.med.jhu.edu (Robert Lewis) (02/20/91)
You ought to check out a package called apE. You can get information about it by ftp at apE.osgp.osc.edu (128.146.18.18) or e-mail to apE@apE.osgp.osc.edu. -- ------------------------------------------------------------------------------ Rob Lewis lewis@cortex.med.jhu.edu Pearlson Neuroimaging Lab (301) 955-8492 Johns Hopkins School of Medicine