PO29

Abstract

A novel mobile cone-beam computed tomography (CBCT) imaging device (Elekta Inc.) aims to improve efficiency with on-demand imaging in the HDR vault during and after applicator insertion. We evaluated whether inter-observer contour variability of standard organs-at-risk (OARs) was similar using this novel CBCT (Ring-CBCT) as compared to diagnostic-quality simulation CT (CT-Sim) for cervical cancer HDR brachytherapy. A patient with stage IIB cervical cancer was implanted with the CT/MR Venezia Tandem (6-cm, 30-deg) with 26-mm dia. Lunar Ovoids. Images were acquired with Ring-CBCT (Elekta ImagingRing v 2.5.2) using 120 kVp, 8 mA, and 0.6x0.6x1.2 mm voxel size, and CT-Sim (Siemens SOMATOM go.Open Pro) using department standard pelvis imaging protocol (120 kVp, 1.17x1.17x1 mm voxel size, 0.8 pitch). Zypher transfer system (Orfit Inc.) was used to move patient to reduce applicator motion. Bladder, rectum, sigmoid, and lower bowel were independently contoured by 11 brachytherapists on Ring-CBCT and CT-Sim in Oncentra TPS. After each session, the physician answered a survey with a score from 1-10 for a qualitative evaluation of their comfort level for each contour. Inter-observer contour variability was quantitatively evaluated using Average Surface Distance, 95% Housdorff Distance, 100% Housdorff Distance, Surface Overlap, Surface Dice, and Volumetric Dice comparisons between a reference contour (RA) and each physician's contour. Wilcoxon signed-rank test was applied to test the statistical difference between the metrics on CT-sim and Ring-CBCT. Eleven physicians completed all four OAR contours on both imaging modalities and completed all surveys. The questionnaire results revealed there was greater confidence in the CT contours when compared to the Ring-CBCT contours. Comfort levels were a median of 10 (9-10) vs. 8 (6-10 (p<0.01) for bladder, 9 (8-10) vs. 8 (4-10) (p<0.01) for rectum, 8.5 (8-10) vs. 6.5 (2-10) (p<0.001) for sigmoid and 8 (7-10) vs. 5 (1-10) (p<0.001) for bowel, respectively. Inter-observer variations in OAR delineation on CT-sim and Ring CBCT are shown in Table 1. Overall, CT-sim showed lower interobserver variation when compared to the Ring-CBCT. There were no statistically significant differences between contour variability with Ring-CBCT when compared to contour variability assessed with CT-sim for most OARs and metrics evaluated. Bladder showed the most prevalent statistically significant variability differences between CT-Sim and Ring-CBCT as evaluated by surface metrics (Average Surface Distance and Hausdorff Distance). Using a multi-center approach and a comprehensive suite of comparison metrics, this study provides the first report of image quality assessment of a novel Ring CBCT for HDR applications. Our data show that for the presented patient, the contours generated using the Ring-CBCT show similar variability when compared to contours generated using standard of care CT-Sim imaging. Based on the physician survey however, OARs contouring confidence with CT-Sim is higher than Ring-CBCT. This study provides a framework for future studies of CBCT imaging for OARs delineation and treatment planning for gynecological cancer brachytherapy.