Title: Deformable Contour Propagation of Organs at Risk in Adaptive Head and Neck Radiation Therapy
Abstract: To quantify the accuracy of deformable image registration and automatic re-contouring of organs-at-risk for head and neck cancers for potential clinical use in adaptive radiation therapy (ART). This retrospective analysis incorporated all head and neck cancer patients who underwent adaptive re-planning between August 2011 and January 2013. Twelve patients went through an initial planning CT and one additional planning CT during their radiation therapy course for adaptive re-planning. Circumstances for re-planning were due to changes in the patient's anatomy during the course of treatment, such as weight loss or tumor response. Two organs-at-risk, the parotid gland and the eye, were contoured on the initial planning CT and the re-planning CT by the same person. Subsequently, a multi-modality image registration and contouring software was used to automatically and deformably propagate contours from the initial planning CT scan to the re-planning CT scan. The software-propagated contours were then compared with the physician-drawn contours using the Dice Similarity Coefficient (DSC). Six males and six females with head and neck cancer were included in this study. The mean DSC of all combined contours was 0.903 with a standard deviation of 0.024. The mean DSCs for the eye were 0.929 and 0.926 for left and right eye, respectively. The mean DSCs for the parotid gland were 0.876 and 0.881 for left and right, respectively. According to previous studies in the literature, DSC value greater than 0.700 suggests a good overlap. The results demonstrated reliable parotid and eye auto-contour propagations from the initial planning CT to the re-planning CT using this type of software. This suggests potential time-savings as this may obviate the need for the physician to re-contour normal organs. A larger patient database and additional organs-at-risk would be required to confirm the findings of this study.