Abstract: We describe a computational model of cancer metastasis that uses an anatomy ontology, the University of Washington Foundational Model of Anatomy, together with a simple theory of tumor spread through the lymphatic system, enabling the automatic calculation of clinical target volumes for radiation therapy planning. It has long been believed that the main mechanism of spread of tumor cells in the head and neck is through the lymphatics. Some involved nodes are visible or palpable, but often the disease spread is microscopic and its extent can only be guessed based on clinical experience. This is the basis for drawing (by hand) clinical target volumes, defining the lymph nodes to treat for specific tumors. However, elementary logic implemented in simple recursive functions together with a probability table can in principle predict patterns of spread. Once the lymph nodes at risk for microscopic disease are known, the next computational challenge is to automatically generate the contours that enclose these nodes. We are making progress on both steps but a system that is ready for regular use in a cancer clinic is still a little further in the future. This project is an example of how a new kind of "theoretical medicine" can be built on logic formalisms. It is joint work with Linda Shapiro (CSE), Mark Whipple (Head and Neck Surgery), various contributors from Radiation Oncology, Jim Brinkley (BHI) and students from the CSE, EE and BHI graduate programs.
The talk will feature a live demo of the actual tumor spread code and if time allows, the Prism radiotherapy planning system.