Program Information
Forecasting Longitudinal Changes in Oropharyngeal Tumor Morphology Throughout the Course of Head and Neck Radiation Therapy
A Yock1,2*, A Rao1, L Dong3, B Beadle1, A Garden1, R Kudchadker1, L Court1, (1) UT MD Anderson Cancer Center, Houston, TX, (2) UT Graduate School of Biomedical Sciences, Houston, TX, (3) Scripps Proton Therapy Center, San Diego, CA
Presentations
MO-C-17A-4 Monday 10:15AM - 12:15PM Room: 17APurpose:
To generate, evaluate, and compare models that predict longitudinal changes in tumor morphology throughout the course of radiation therapy.
Methods:
Two morphology feature vectors were used to describe the size, shape, and position of 35 oropharyngeal GTVs at each treatment fraction during intensity-modulated radiation therapy. The feature vectors comprised the coordinates of the GTV centroids and one of two shape descriptors. One shape descriptor was based on radial distances between the GTV centroid and 614 GTV surface landmarks. The other was based on a spherical harmonic decomposition of these distances. Feature vectors over the course of therapy were described using static, linear, and mean models. The error of these models in forecasting GTV morphology was evaluated with leave-one-out cross-validation, and their accuracy was compared using Wilcoxon signed-rank tests. The effect of adjusting model parameters at 1, 2, 3, or 5 time points (adjustment points) was also evaluated.
Results:
The addition of a single adjustment point to the static model decreased the median error in forecasting the position of GTV surface landmarks by 1.2 mm (p<0.001). Additional adjustment points further decreased forecast error by about 0.4 mm each. The linear model decreased forecast error compared to the static model for feature vectors based on both shape descriptors (0.2 mm), while the mean model did so only for those based on the inter-landmark distances (0.2 mm). The decrease in forecast error due to adding adjustment points was greater than that due to model selection. Both effects diminished with subsequent adjustment points.
Conclusion:
Models of tumor morphology that include information from prior patients and/or prior treatment fractions are able to predict the tumor surface at each treatment fraction during radiation therapy. The predicted tumor morphology can be compared with patient anatomy or dose distributions, opening the possibility of anticipatory re-planning.
Funding Support, Disclosures, and Conflict of Interest: American Legion Auxiliary Fellowship The University of Texas Graduate School of Biomedical Sciences at Houston
Contact Email: