[CCoE Notice] Ph.D. Dissertation Presentation

[Khator, Suresh]

Deformation Prediction and Geometrical Modeling of Head and Neck Cancer Tumor: A Data Mining Approach
Maryam Azimi, Ph.D. Candidate
Advisor: Dr. Ali Kamrani
Co-Advisor: Dr. Marvin Karson

April 20th 2011, 2:30- 4:30 pm
Industrial Engineering Conference Room


Abstract
     Radiation therapy has been used in treatment of cancer tumors for several years. It may be used as primary therapy or with combination of surgery or other kinds of therapy such as chemotherapy and hormone therapy. The treatment objective is to destroy cancer cells or shrink the tumor by planning an adequate radiation dose to the desired target without damaging the normal tissues. By using the pre-treatment Computer Tomography (CT) images, most of the radiotherapy planning systems design the target and assume that the size of the tumor will not change throughout the treatment course, which takes 5 to 7 weeks. Based on this assumption, the total amount of radiation is planned and fractionated for the daily dose required to be delivered to patient's body. However, this assumption is flawed because the patients receiving radiotherapy have marked changes in tumor geometry during the treatment period. Therefore, there is a critical need to understand the changes of the tumor shape and size over time during the course of radiotherapy in order to prevent significant effects of inaccuracy in the planning.
     In this research, a methodology is proposed in order to monitor and predict daily (fraction day) tumor volume and surface changes of head and neck cancer tumors during the entire treatment period. In the proposed method, geometrical modeling and data mining techniques will be used rather than repetitive CT scans data to predict the tumor deformation for radiation planning. Clinical patient data are obtained from the University of Texas-MD Anderson Cancer Center (MDACC). In the first step, by using CT scan data, the tumor's progressive geometric changes during the treatment period are quantified. The next step relates to using proper data mining techniques in order to develop predictive models for tumor geometry based on the patients' selected attributes (age, weight, stage, etc.). Moreover, statistical analyses have been applied to identify the effects of patients' selected attributes on tumor deformation. The main goal of the proposed methodology is increasing the accuracy of each therapy and quality of life for patients.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://Bug.EGR.UH.EDU/pipermail/engi-dist/attachments/20110404/560ae3bb/attachment.html