This data was submitted by Tom Palmer ( palmer@ncsc.org )

Enjoy -

Terry

---
Terry Myerson
International AVS Center
North Carolina Supercomputing Center
tvv@ncsc.org

-----------------------------------

			     DESCRIPTION

Multichannel cardiac mapping systems are used to acquire, analyze, and
display electrical data sensed by electrodes placed directly on or
within the heart tissue.  The electrodes measure the potentials
generated by biological electrical sources within the heart tissue and
those due to externally applied electrical fields from defibrillators
and pacemakers.  The cardiac potentials are analyzed to determine the
approximate times that activation wavefronts passed beneath or near
the electrode.  After a study, the heart of the experimental animal
(in this case, a dog) is removed and fixed in formalin.  The
acquisition of anatomic information is then carried out using MRI.

The information gathered as described is used to compute secondary
variables, interpolate values of sparsely measured variables, and
display the results.  The data are traditionally displayed as contour
lines superimposed on stylized or individualized cardiac contours.
However, the volumetric nature of the data, and the need to view
multiple variables simultaneously, lead to attempts to use
multivariate volume visualization techniques to display the data.
This dataset was the impetus for the development of the language-based
multivariate classification and shading techniques described in:

T.C. Palmer, E.V. Simpson, K.M. Kavanagh, and W.M. Smith.
"Visualization of Bioelectric Phenomena".  In "High Performance
Computing in Biomedical Research".  CRC Press, October 1992.


				 DATA

The dataset actually consists of five separate (but registered)
volumes.  All volumes are 128x128x128 with one byte per voxel.  Z
varies fastest.  Splice them together as required.

    act.raw		Activation time (time at which the wavefront
			passed this point).

    pot.raw		Defibrillation shock potential

    pot-mag.raw		Defibrillation shock potential gradient
			magnitude

    valid.raw		Valid data region (since the bioelectric data
			is only acquired or interpolated for voxels
			inside the myocardium).  Encoding:
				undefined:	0
				boundary:	128
				defined:	255

    mri.raw		MRI scan of canine heart


			     EXPERIMENTS

Try extracting isosurfaces of activation to display in conjunction
with MRI surfaces.  This yields an "isochrone" - a visual depiction of
the wavefront.  We've colored these surfaces by several variables
including conduction velocity.  Since the gradient of the activation
data represents the change in time with respect to distance,
1/|gradient| serves as an approximation of conduction velocity.

Other things to try:

	Coloring isochrones by shock potential gradient magnitude.
	Coloring the MRI surface by shock potential.


			       CREDITS

The MR dataset was acquired in the microscopic MRI laboratory of G.
Allan Johnson with technical assistance by Sharon Melnick (Duke
University).  The bioelectric data was acquired by Katherine Kavanagh
(University of Alberta).  Interpolation and registration were
performed by Ed Simpson (Duke).  Visualization work was done by Tom
Palmer (Cray Research), Ed Simpson, and Bill Smith (Duke).


			       CONTACT

Tom Palmer						(919) 248-1117
Cray Research, Inc.					palmer@ncsc.org
North Carolina Supercomputing Center
PO BOX 12889
RTP, NC 27709