Researchers of the Mednet Laboratory at the Medical Faculty of the Goteborg University have developed an educational software package, called 3D-Brain 2.0, which focuses on the anatomy and functionality of the human brain. Their major concern, expressed at the "Medicine Meets Virtual Reality" (MMVR) Conference in San Diego, last winter, was to create a platform independent learning and training tool for students, that is able to run on a basic personal computer with use of the latest visualisation technologies such as the CD-ROM based QuickTime VR 2.0 and VRML 2.0. Before the MMVR-audience, they offered a methodological and target-based comparison between the graphical capacities on both a high end workstation and a standard personal computer.
In Sweden, just like elsewhere, funds for educational equipment are drastically being cut back. Teachers and researchers are forced to develop their own didactic medical software while being less experienced in high skill programming than the expensive real professionals and software companies. Yet, medical students should learn to work with recent scientific visualisation techniques and analytical instruments because IT knowledge and computer literacy are indispensable for the future generation of health care providers. Today, the soft- and hardware performance of the personal computer almost matches the standard of high end workstations. As a result, the Mednet research team has generated a PC-based visualisation package, derived from 3D datasets, which have been manipulated on Silicon Graphics computers.
The 3D-Brain 2.0 programme is based on human brains, donated to the Medical Faculty in Goteborg. The researchers preferred rather not to use the Visible Human Datasets from the National Library of Medicine, since they didn't like the idea of a "prototypical human being" that unavoidably would emerge if everyone utilised the same material. The brains were sliced in 1 mm sections, photographed, and digitised. The final datasets were obtained by manual tracing, splines triangulation, and 3D visualisation. Subsequently, the VRML code was generated from the original datasets for the VRML 2.0 version. To produce the QuickTime VR (QTVR) package, 684 images with a 10∞ angle were created for each scene and introduced to an Apple Macintosh computer for further processing. All interactive features were executed on a Macintosh and afterwards integrated in the Windows95 PC platform.
In order to select the best file format and platform for visualisation, both target group and producer of the didactic tool have to be determined. Medical students need CD-ROM or Web-based applications, since they perform their research in resource centres or at home. Medical researchers or teachers usually are not familiar with expert programming, so they want a relatively simple platform to work from. In this context, Open Inventor, the prevailing file format standard on high end workstations, offers superior texture mapping abilities at an incredible speed, but its SceneViewer is less suitable for pedagogic purposes. Working with SceneViewer indeed requires too complicated a programming effort and moreover, this sophisticated application is only available on expensive high end computers.
Fortunately, VRML and QTVR constitute two affordable alternatives for implementation of the 3D-Brain 2.0 package. The platform independent VRML general didactic structure is written in HTML and can be combined with frames and Java. The major disadvantage forms its lack of speed in complex visualisation applications. In turn, QTVR is based on photographic or high quality rendered images instead of on wireframe models. The user is able to zoom an object, define hotspots and produce overlays. Speed entirely depends on disc reading capacity, as QTVR is CD-ROM based. The didactic framework is provided by Macromedia Director 6.0. Possible drawbacks might include the limited zoom and pan capacity while file sizes are 10 to 30 times that of a VRML or SceneViewer file.
Comparison by the Mednet research team shows the usability of all three 3D visualisation formats but in different and specific circumstances. Scientific research is best performed with SceneViewer, which equally provides the basis for generating the graphic material, used in both VRML and QTVR settings. The latter two ideally serve as perfect educational tools, even on simple standard personal computers. A complete comparison report is provided at the Mednet site of the Medical Faculty of Goteborg University.