The Visible Human within the VOXEL-MAN Framework
Thomas
Schiemann,
Ulf Tiede,
Karl Heinz
Höhne
Institute of Mathematics and Computer Science in Medicine (IMDM),
University Hospital Hamburg-Eppendorf, Germany
e-mail: schiemann@uke.uni-hamburg.de
WWW : http://www.uke.uni-hamburg.de/idv
In previous work we have developed a framework for generating
volume based 3D
interactive atlases from cross-sectional images
[4,
7,
10,
14,
15,
16,
18]. These
atlases are based on a two layer model of image volumes linked to
a semantic
network containing descriptive knowledge (see Figure 1).
For extraction of the model's contents there is a large set of
visualization,
exploration, and simulation tools. The pictorial basis of
existing VOXEL-MAN
atlases are radiologic cross-sectional images. While this has the
advantage
that anatomy can be linked to radiological imaging, anatomical
detail is
subject to improvement. The high resolution data sets of the
Visible Human
project
[17]
are therefore an ideal basis for
this purpose.
Typically the Visible Human images are presented in stacks of
orthogonal
slices, which can be browsed through. Compared to such simple
systems,
creation of a real 3D model requires detailed segmentation, which
is a huge
project due to the vast amount of data and anatomical detail. In
our first
approach we overcome this problem by registration of the Visible
Human with
the existing volume based atlas VOXEL-MAN/brain. This procedure
introduces
anatomical detail from the Visible Human into the radiological
environment of
VOXEL-MAN/brain, and on the other hand the detailed labelling of
VOXEL-MAN can
be correlated with the Visible Human [1]. This procedure
[12] adds a new
functionality to
the existing atlas, 3D
display of the Visible Human is not possible in this way.
The real challenge is, however, the construction of a volume
model of the
Visible Human, which allows the interactive exploration and the
derivation of
labelled pictures and animations with realistic visualization.
The VOXEL-MAN
environment is ready for this. Yet we had to adapt segmentation
and rendering
to the multi-parametric nature of the data. For segmentation we
extended the
threshold based methods developed for scalar data
[6, 11]. For the color images the
intensity
threshold range defining an object was replaced by an ellipsoid
in RGB-space
[13]. The volume
visualization
algorithms developed previously
[2,
3,
5,
8,
9
19] where
modified in that the classification defined by the ellipsoid is
used for
supersampling in the ray-tracing step [20]. The surface colors
are texture mapped from the photographic data.
We produced experimental atlases of the Visible Human's head (16
Mvoxels),
shoulder (35 Mvoxels), and abdomen (40 Mvoxels). We reduced the
in-slice
resolution to 1 mm (yielding 1 cmm voxel-size) in order to keep
the models in
reasonable sizes. We segmented only the major anatomical
components so far. It
turns out that interactive use of the thus generated atlases is
possible at
reduced image quality (10-15 seconds per view). The nearly
photorealsistic
quality achievable with the above described methods must be paid
for with
longer computation times. For instance
Figure 7 with
1254 x 983 pixels took 45 minutes on a DECstation 250 with 266
MHz. To
overcome this problem we can precompute QuickTime-VR movies,
which allow
interaction with 2 degrees of freedom.
Further information and some recent papers are found under the
URL
http://www.uke.uni-hamburg.de/Institutes/IMDM/IDV/
VisibleHuman.html
This paper concentrates on the additions made to VOXEL-MAN for
incorporating
the Visible Human data.
Yet VOXEL-MAN is the result of a project to which many
co-workers have contributed.
Andreas Pommert and Rainer Schubert have designed and implemented
the knowledge
base and its user interface.
Martin Riemer is in charge of the general user
interface to which also Bernhard Pflesser and Kay Priesmeyer
contributed.
Hans-Christian Wulf helped us by generating and
providing tools for the movies.
Kay Priesmeyer created the HTML version of this paper.
- 1
-
Höhne, K. H. (Ed.): VOXEL-MAN, Part 1: Brain and Skull,
Version 1.1.
Springer-Verlag Electronic Media, Heidelberg, 1996.
(ISBN 3-540-14584-2).
- 2
-
Höhne, K. H., Bernstein, R.: Shading 3D-images from CT using
gray level
gradients.
IEEE Trans. Med. Imaging MI-5, 1 (1986), 45-47.
- 3
-
Höhne, K. H., Bomans, M., Pommert, A., Riemer, M., Schiers,
C., Tiede, U.,
Wiebecke, G.: 3D-visualization of tomographic volume data using
the
generalized voxel-model.
Visual Comput. 6, 1 (1990), 28-36.
- 4
-
Höhne, K. H., Bomans, M., Riemer, M., Schubert, R., Tiede,
U., Lierse, W.: A
3D anatomical atlas based on a volume model.
IEEE Comput. Graphics Appl. 12, 4 (1992), 72-78.
- 5
-
Höhne, K. H., DeLaPaz, R. L., Bernstein, R., Taylor, R. C.:
Combined surface
display and reformatting for the 3D-analysis of tomographic
data.
Invest. Radiol. 22 (1987), 658-664.
- 6
-
Höhne, K. H., Hanson, W. A.: Interactive 3D-segmentation of
MRI and CT
volumes using morphological operations.
J. Comput. Assist. Tomogr. 16, 2 (1992), 285-294.
- 7
-
Höhne, K. H., Pflesser, B., Pommert, A., Riemer, M.,
Schiemann, T., Schubert,
R., Tiede, U.: A new representation of knowledge concerning
human anatomy and
function.
Nature Med. 1, 6 (1995), 506-511.
- 8
-
Höhne, K. H., Riemer, M., Tiede, U.: Viewing operations for
3D-tomographic
gray level data.
In Lemke, H. U. et al. (Eds.): Computer Assisted Radiology,
Proc. CAR '87, Springer-Verlag, Berlin, 1987, 599-609.
- 9
-
Pommert A., Bomans M., Höhne K. H.: Volume visualization in
magnetic
resonance angiography.
IEEE Comput. Graphics Appl. 12,
4, (1992), 12-13
- 10
-
Pommert, A., Schubert, R., Riemer, M., Schiemann, T., Tiede, U.,
Höhne,
K. H.: Symbolic modeling of human anatomy for visualization and
simulation.
In Robb, R. A. (Ed.): Visualization in Biomedical Computing
1994, Proc. SPIE 2359.
Rochester, MN, 1994, 412-423.
- 11
-
Schiemann, T., Bomans, M., Tiede, U., Höhne, K. H.:
Interactive
3D-segmentation.
In Robb, R. A. (Ed.): Visualization in Biomedical Computing
II,
Proc. SPIE 1808.
Chapel Hill, NC, 1992, 376-383.
- 12
-
Schiemann, T., Höhne, K. H., Koch, C., Pommert, A., Riemer,
M., Schubert, R.,
Tiede, U.: Interpretation of tomographic images using automatic
atlas lookup.
In Robb, R. A. (Ed.): Visualization in Biomedical Computing
1994, Proc. SPIE 2359.
Rochester, MN, 1994, 457-465.
- 13
-
Schiemann, T., Nuthmann, J., Tiede, U., Höhne, K. H.:
Segmentation of the
Visible Human for high quality volume based visualization.
In Höhne, K. H., Kikinis, R. (Eds.): Visualization in
Biomedical Computing, Proc. VBC '96, Lecture Notes in
Computer Science
1131, Springer-Verlag, Berlin, 1996, 13-22.
- 14
-
Schubert, R., Höhne, K. H., Pommert, A., Riemer, M.,
Schiemann, T., Tiede,
U., Lierse, W.: A new method for practicing exploration,
dissection, and
simulation with a complete computerized three-dimensional model
of the brain
and skull.
Acta Anat. 150, 1 (1994), 69-74.
- 15
-
Schubert R., Höhne K. H., Pommert A., Riemer M., Schiemann
T.,
and Tiede U.: Spatial knowledge representation for
visualization of
human anatomy and function. In Barrett H. H. and Gmitro A. F.
(Eds.):
Information Processing in Medical Imaging, Proc. IPMI '93,
Lecture Notes in Computer Science 687, Springer-Verlag, Berlin,
(1993), 168-181
- 16
-
Schubert R., Pommert A., Höhne K. H.: Methods for
model-based
knowledge representation in anatomy.
Annu. Symp. Comput. Appl. Med. Care Proc. SCAMC,
New Orleans, LA, (1995), 971 (abstract)
- 17
-
Spitzer, V., Ackerman, M. J., Scherzinger, A. L., Whitlock, D.:
The Visible
Human Male: A Technical Report.
J. Am. Med. Inf. Ass. 3, 2 (1996), 118-130.
- 18
-
Tiede, U., Bomans, M., Höhne, K. H., Pommert, A., Riemer,
M., Schiemann, T.,
Schubert, R., Lierse, W.: A computerized three-dimensional
atlas of the human
skull and brain.
Am. J. Neuroradiology 14, 3 (1993), 551-559.
- 19
-
Tiede, U., Höhne, K. H., Bomans, M., Pommert, A., Riemer,
M., Wiebecke, G.:
Investigation of medical 3D-rendering algorithms.
IEEE Comput. Graphics Appl. 10, 2 (1990), 41-53.
- 20
-
Tiede, U., Schiemann, T., Höhne, K. H.: Visualizing the
Visible Human.
IEEE Comput. Graphics Appl. 16, 1 (1996), 7-9.
Figure 1:
Basic framework of the VOXEL-MAN atlases.
Figure 2:
Reconstruction of muscular and vascular structures of the head
and a
magnification of the center portion.
Figure 3:
Three illustrations of VOXEL-MAN's ability to cut, map and view different modalities in one rendering.
Figure 4:
Two cut plane views of the abdomen.
Figure 5:
A view into the Visible Human's stomach and a picture of a real gastroscopic
image.
Figure 6:
User interface of the experimental atlas VOXEL-MAN/VHP-abdomen.
Figure 7:
Two realistic views: the right shoulder with bones, musculature and some major
veins; and the torso.
Figure 8:
Three red/green stereo views.
Movies:
Two QuickTime VR movies.