Department of Computer Science
			    University of Utah
		       Salt Lake City, UT 84112 USA

			     October 26, 1994

		 Construction of a human torso model from
		 magnetic resonance images for problems in
		    computational electrocardiography.

		Robert S. MacLeod (macleod@cvrti.utah.edu)
		 Phil R. Ershler (ershler@cvrti.utah.edu)
		 Christopher R. Johnson (crj@cs.utah.edu)

				UUCS-94-017

				 Abstract


Applying mathematical models to real situations often requires the use of
discrete geometrical models of the solution domain.  In some cases
destructive measurement of the objects under examination is acceptable, but
in biomedical applications the measurements come from imaging techniques
such as X-ray, computer tomography (CT), or magnetic resonance imagining
(MRI).  A necessary early step in the modeling process is then to extract
from these images the measurements (locations and distances) that form the
basis of the geometrical model.

In this paper we describe the construction of a geometrical model of the
human thorax based on the high resolution MRI scan of a single subject.  We
outline the scanning procedure, the image collection and conversion to
computerized image files, the segmentation of the images into boundary
nodes, and the connection of these nodes into surface, and then volume,
meshes.  Included are brief descriptions of the tools developed at the
CVRTI for this project, as well as our experiences in creating and using
them.  The result of this work was a pair of models at two different levels
of spatial resolution, which set new standards in the area of bioelectric
field modeling and the application of these models has been described
previously.