This project was part of an honours thesis at the Computer Science Department, Australian National University. It made use of the digital archive of state of the art radiographic and photographic images from the Visible Human Project (VHP), National Library of Medicine (NLM), Maryland. A brief description of the scenario is given here.

The format of the radiological and photographic images being non-standard, conversion algorithms were required to port them to local platforms made up of SUN workstations, PCs, Macintoshes and the Fujitsu AP1000. The converted data, available as a set of portable graymaps (PGM) and pixmaps (PPM) was then visualised on the SUN, PC and Macintosh platforms as two dimensional images using both existing browsing software and new ones. Here are a few sample images of the Computed Tomography Scans, Magnetic Resonance Images and Photographic Images.

A registration model was applied to the data set, using the fresh CT scans and the colour photographs, making use of the features within the data to serve as registration marks. Having isolated several features within the photographic images, by means of a segmentation process, those features were used to reconstruct volume data made up of several planes of the photographic images. The reconstruction process had to take into account the format of the volume data required by the renderer, vol (serial) and cap_vol (parallel), a visualisation software developed at the Computer Science Department, Australian National University (ANU).

The volume data was then ported onto the cells of the AP1000 supercomputer and rendered using various rendering options. Optimisation techniques were devised to make the rendering process efficient and promising results were obtained from the project in that respect and as a whole. A few snapshots of sample volumes representing different features of the Visible Human photographic image archive are shown here. The results proved to be beneficial towards achieving the goal of the Visible Human Project which is to produce a system of knowledge structures which will transparently link visual knowledge acquired from this project to symbolic knowledge formats such as theories, hierarchies, names and principles.

You may view the entire thesis online or download a compressed postscript version (8 Mb).