The Clinical Target Volume (CTV) project
The Clinical Target Volume project has as its objective to create a
mathematical model of how tumors spread to regional lymph nodes and
other locations from the primary tumor. It is a collaborative project
of Radiation Oncologists, Otolaryngologists, and Computer Scientists.
The basis for the model is that local and regional tumor dissemination
is generally believed to take place through the lymphatic system,
starting with the lymphatic vessels that drain the anatomic site of
the the primary tumor. The University of Washington Foundational
Model of Anatomy
contains a computationally
accessible representation of the lymphatic system and its components.
This representation can be used to generate the topology of the tumor
spread model. The model itself is a Markov chain.
We now also have a secondary goal, of generalizing the method for
creating such models from other ontological knowledge representations,
besides the FMA.
Initial work on this idea examined the use of a rule-based
representation to predict involved lymph nodes. This was then revised
to represent lymphatic vessels as a connected set of Markov chains,
one for each lymphatic vessel in a path. We are now working on a more
unified Markov chain representation with fewer free parameters.
Publications related to the CTV model project
- Kalet, I.J., Whipple, M., Pessah, S., Barker, J.,
Austin-Seymour, M.M., Shapiro, L.G. A Rule-based Model for Local
and Regional Tumor Spread. Proceedings of the American Medical
Informatics Association (AMIA) Fall Symposium, pp. 360--364, Isaac
S. Kohane, ed., Hanley and Belfus, Inc., 2002.
- Benson, N., Whipple, M. and Kalet, I.J. A Markov Model Approach
to Predicting Regional Tumor Spread in the Lymphatic System of the
Head and Neck, Proceedings of the American Medical Informatics
Association Fall Symposium, pp. 31--35, David W. Bates, John
H. Holmes and Gilad Kuperman, Eds., American Medical Informatics
Association, 2006. PDF
Once the probability of tumor presence in various lymph nodes is
computed, and a threshold is chosen, there is still the problem of
drawing the actual contours of a target for radiotherapy planning,
which will include the involved nodes but as little else as possible.
This problem can be addressed using an atlas of images in which the
various lymph node groups are drawn, together with image registration
techniques to map the standard atlas to the particular anatomy of an
individual patient. This part of the project is not currently
Publications related to the contour generation project
- Teng, C.C., Austin-Seymour, M.M., Barker, J., Kalet, I.J.,
Shapiro, L.G., Whipple, M. Head and Neck Lymph Node Region
Delineation with 3-D CT Image Registration. Proceedings of the
American Medical Informatics Association (AMIA) Fall Symposium,
pp. 767--771, Isaac S. Kohane, ed., Hanley \& Belfus, Inc., 2002.
- Teng, C.C., Shapiro, L.G., and Kalet, I.J. Head and Neck Lymph
Node Region Delineation Using a Hybrid Image Registration Method.
Proceedings of the IEEE International Symposium on Biomedical
Imaging, 2006, pp. 462-465, IEEE Computer Society, 2006.
- Teng, C.C., Shapiro, L.G., and Kalet, I.J. Automatic
Segmentation of Neck CT images. Proceedings of the 19th IEEE
International Symposium on Computer-based Medical Systems,
pp. 442-445, D.J. Lee, Brian Nutter, Sameer Antani, Sunanda Mitra,
and James Archibald, Eds., IEEE Computer Society, 2006.
- Teng, C.C., Shapiro, L.G., and Kalet, I.J. Head and Neck Cancer
Patient Similarity Base on Anatomical Structural Geometry.
Proceedings of the 2007 IEEE International Symposium on Biomedical
Imaging (ISBI): From Nano to Macro, in press, IEEE Engineering in
Medicine and Biology (EMB) Society, 2007.