1273 - Prospective Evaluation of a C-Arm Based Advanced On-Board Cone-Beam Computed Tomography (CBCT) System for Treatment Planning Feasibility across Multiple Anatomic Sites
Presenter(s)
S. Lichtman-Mikol1, A. T. Price1, T. Arsenault1, J. A. Dorth2, K. Gregg2, B. Guevara2, A. Y. Jia1, R. Kashani1, Q. Li1, P. Mohindra1, R. K. Patel1, H. K. Perlow1, C. Petersen1, A. Rezaei1, D. E. Spratt3, E. A. Wiegner4, R. Woods5, M. Xing6, and L. E. Henke2; 1Department of Radiation Oncology, University Hospitals Cleveland Medical Center/ Seidman Cancer Center, Cleveland, OH, 2Department of Radiation Oncology, University Hospitals Seidman Cancer Center, Case Western Reserve University, Cleveland, OH, 3University Hospitals Seidman Cancer Center, Case Western Reserve University, Cleveland, OH, 4Mercy Medical Group, Sacramento, CA, United States, 5University Hospitals, Seidman Cancer Center, Case Western Reserve University, Cleveland, OH, 6Case Western Reserve University/University Hospitals Cleveland Medical Center, Cleveland, OH
Purpose/Objective(s): Standard C-arm based cone-beam computed tomography (CBCT) imaging for image-guided radiotherapy (IGRT) has limited soft tissue visualization and contrast resolution for treatment planning and adaptive RT. A novel, advanced CBCT system (HyperSight; hCBCT) was previously demonstrated to approach diagnostic CT quality on a ring gantry platform but has not been similarly evaluated on a C-arm platform. We hypothesized that hCBCT image quality would be non-inferior to CT simulation (TPCT) for RT planning across multiple anatomic sites, assessed by organ-at-risk (OAR) delineation and Hounsfield Unit (HU) accuracy.
Materials/Methods: Forty patients receiving IGRT were enrolled in a prospective imaging trial across four cohorts: results are reported here for abdomen and pelvis cohorts (n=20; NCT05975619). Feasibility was pre-defined as ³80% of hCBCT scans meeting non-inferiority thresholds for both OAR contour agreement and HU dose recalculation accuracy. Pelvic OAR were bladder, lower-GI (sigmoid colon, rectum); the abdominal OAR is total GI (stomach, duodenum, small bowel, large bowel). Nine raters (physicians and medical physicists) delineated OAR on TPCT and hCBCT image pairs; a Simultaneous Truth and Performance Level Estimation (STAPLE) consensus contour served as reference on each. Pairwise agreement was quantified using Dice Similarity Coefficient (DSC) and Average Symmetric Surface Distance (ASSD), with non-inferiority margins of <2 mm and <5%, respectively. Contouring confidence was rated by Likert scale. HU accuracy was assessed by recalculating clinical plans on hCBCT and comparing to TPCT-based plans via 3D gamma analysis (1%/1mm) and dose-volume histogram (DVH) target structure metrics (D99%, Dmean, D1%).
Results: Twenty patients (20 TPCT, 20 hCBCT) were contoured by 9 raters. Feasibility for RT planning was met for pelvis (10/10) but not abdomen (5/10) hCBCTs (15/20 total feasible) by contour criteria. On contour analysis, overall mean DSCs were 0.89±0.07 on hCBCT vs 0.91±0.04 on TPCT, with mean ASSD of 1.94±1.11 mm and 1.78±1.52 mm, respectively. For abdomen hCBCT vs. TPCT, mean DSC was 0.84±0.10 vs 0.91±0.03 and mean ASSD was 2.33±1.39 mm vs 1.42±0.56 mm, respectively. In contrast, pelvis hCBCT had comparable DSC (0.905±0.049 vs 0.905±0.045) and ASSD (1.77±0.92 mm vs 1.95±1.76 mm) to TPCT. Rater confidence was significantly lower for hCBCT vs TPCT (median 4 vs 5, p<0.001), most notably in abdomen (median 3 vs 5, p<0.001), with rater comments citing bowel gas artifacts. Dose recalculation on hCBCT yielded 3D gamma pass rates of 96.2±3.6% (10% threshold) and 95.0±6.9% (50% threshold) at 1%/1mm, with <1% DVH target metric difference from TPCT, establishing HU accuracy for 20/20 cases.
Conclusion: Use of hCBCT is feasible for pelvic RT planning based on OAR delineation and HU accuracy. In abdomen, HU accuracy was maintained for all cases, but inter-rater contour variability precluded meeting the 80% threshold (5/10). Analysis of additional disease sites is ongoing.