The Visible Human Project (VHP) of the National Library of Medicine has created the world's most detailed visual description of the human body. However, the educational value of the database is limited by the fact that for a given point in space, the color is known, but not the identity of the anatomic structure. The project in which the database is "segmented and classified" is currently being carried out by Dr. Victor Spitzer, the original principal investigator of VHP, in cooperation with Gold Standard Multimedia Inc. This will enable the next generation of anatomy educational products, which will begin to take on a "virtual reality" feel. [See http://w1.uchsc.edu/sm/chs/segclass.html for more information.]
The process of "segmentation" involves distinguishing one anatomic structure from another based on visual or numerical interpretation of the data. It is carried out "slice by slice" on the individual images produced by the original digital photography. Much of this job can be done algorithmically with edge-detection routines not unlike those in common image processing programs. However, the most critical part of the process involves interpretation and manual segmentation by experienced anatomists, i.e. Dr. Spitzer and his colleague, Dr. David Whitlock.
"Classification" refers specifically to relating each segmented item to its correct anatomic name. This process can be difficult when dealing with very small items. Also, careful use of terms relating to anatomic structures and substructures will impact the ease with which the data can be utilized by various software packages.
One of the exciting uses of the Visible Human data is to digitally stack the slices on top of one another to "render" a three-dimensional scene of the body from any angle and with arbitrary environmental parameters (lighting, wall position and color, etc.). By rendering the segmented and classified data, one can easily see the value of this data in identifying the shape and location of individual anatomic structures and the relationships between two or more structures. It also clearly demonstrates the ultimate goal, which is to render the visual data in parallel with the identification data to bring real educational and research value to the student and scientist in the form of a "virtual reality" atlas of anatomy.