SEGMENTATION Segmentation is the process of identifying and describing features within an image or data set. It is a necessary to constructing a higher level description of the contents of the image. The automatic segmentation of features in 2D images has a long history in the fields of artificial intelligence and machine vision. It depends on the ability to rigorously describe the features being sought.

For 3D anatomical data sets such as CT, MRI or the cryosection data of the Visible Human project segmentation is necessary in order to visualize discrete structures. Automatic techniques are suitable for situations where structures can be adequately described in terms of data values (e.g. CT bone versus soft tissue). It is frequently necessary to hand trace features in cross section where there is no adequate description in terms of data values. However, even where automatic methods fail, human experts are able to discern medically useful features within the data. The problem comes in translating the human export knowledge into a description useable by segmentation algorithms. Short of this it is useful to have segmentation algorithms which can be interactively controlled by the user based upon their expert knowledge.

EXAMPLES Stereo pair of hepatic portal system.
Stereo pair of leg muscles.
PRESENTATIONS "An Immersive Environment for the Visualization and Segmentation of Large Volumetric Data Sets", Medicine Meets Virtual Reality VI, San Diego, Jan. 1998.


Last Modified: Tuesday, 13-Apr-99 16:24:44 CDT Server Contact: senger@csfac.uwlax.edu

Copyright © 1995-98, The University of Wisconsin - La Crosse Scientific Visualization Project