About 4D Visualization Research
The focus of my Master's research with Professor David Kaeli was four dimensional (4D) visualization of medical image data sets and anything we can do with them to enhance physicians' ability to understand physical volumes they can't normally see as they move over time. In this context, 4D typically refers to the three spatial dimensions plus some representation of the temporal dimension as the fourth one. For example, because it isn't possible to capture “snapshots” of 3D anatomy instantaneously, four dimensional X-Ray computed tomography (4DCT) data sets are reconstructed by some base assumption that human respiration is more or less periodic and that the subject being imaged will look the same at the same phase of breathing every time.
What Do My Insides Look Like?
It's tough to say what your insides look like, but if you've ever wondered what your lungs, ribs, and skin look like and how they lie relative to each other in your body, then this is probably a good set of examples for you (click for a higher-quality version):
Ski Run?
All images were produced using SCIRun.
“Ski Run” is how you pronounce the name of the problem solving environment we used to do our work: SCIRun.
It is extensible by creating modules, or plugins, that interact with dataflow networks that may exist already (such as the ones used to make the visualizations seen above).
Submitted Modules
I recently submitted more generic versions of my work to the main SCIRun project. Those source files can be found here:
What Were Your Contributions?
We made several modules that could be used in parallel with the built-in SCIRun visualization capabilities to display voxel trajectories (as pre-determined by an image registration algorithm – in this case, deformable registration) simultaneously with anatomy (as displayed above). This was the result:
By itself, the first screenshot does not explain very much. What it shows are several trajectories displayed by dragging a cursor around in the 3D environment. Trajectories are represented by open-ended line loops that start from blue (the first respiratory phase) and make a smooth transition over each of the respiratory phases in the image set to red (the last respiratory phase).
Furthermore, trajectories can be edited. This, clearly, does not mean we can go back into a patient and move cells around to suit our needs 
. However, it does mean that we can observe errors in the results of the two independently processed algorithms (volume rendering and deformable registration), which can have several adjustable parameters, and use our ability to make edits to trajectories interactively as the starting point for adjusting parameters.
In the next two screenshots you can see the trajectory editing tool alone and then superimposed on the trajectory viewing tool.
Publications
Burak Erem, Gregory C. Sharp, Ziji Wu, David R. Kaeli: Interactive Deformable Registration Visualization and Analysis of 4D Computed Tomography. ICMB 2008: 232-239
Presentation
No slides are up yet, but here are the images they apparently took of me while I was presenting.
