Author: Cem Altunbas, Adam Avant, Farhang Bayat, Roy Bliley, Ian Boor, Brian Kavanagh, Uttam Pyakurel, Tyler Robin, Ryan Sabounchi 👨🔬
Affiliation: Department of Radiation Oncology, University of Colorado School of Medicine, Taussig Cancer Center, Cleveland Clinic 🌍
Purpose: Improved soft tissue visualization plays an important role in target localization and treatment plan adaptation in CBCT-guided radiotherapy. In this work, a novel CBCT approach, 2D antiscatter grid-based quantitative CBCT (qCBCT), was implemented, and its effect on tissue delineation accuracy was evaluated in a prospective CBCT imaging trial.
Methods: In an IRB-approved study, nine prostate cancer patients were scanned with the Varian TrueBeam’s iCBCT protocol and the proposed qCBCT approach on the same day and using identical acquisition parameters. Four observers contoured up to seven unique anatomical structures, such as prostate, rectum, bladder, and iliac veins in each CBCT and planning CT (pCT) in a blinded fashion, totaling 468 structure delineations. Inter-observer similarity of contours was evaluated by comparing each observer’s contour to the consensus contour generated via the STAPLE algorithm. Similarity of observer and consensus contours were assessed using Hausdorff distance, mean absolute surface distance (MASD), Dice coefficient, and volume difference. A Wilcoxon Signed-Rank Test was performed to assess the significance of differences of qCBCT and iCBCT.
Results: The average Hausdorff distance for all contours was 6.3±5.2 and 5.4±3.2 mm for iCBCT and qCBCT (p <0.005). Average Dice similarity coefficients were 0.91 and 0.95 for iCBCT and qCBCT (p< 0.005). The average normalized absolute volume differences were 14±13 and 11± 8 % for iCBCT and qCBCT (p< 0.005). qCBCT provided more favorable contour similarity metrics in 2.2 times more anatomical structures than iCBCT. In pairwise comparisons, pCT provided better contour similarity metrics than iCBCT and qCBCT for 67.9% and 57.7% of the structures, respectively.
Conclusion: The 2D antiscatter grid-based qCBCT significantly improved the interobserver agreement in anatomical structure delineations. These results suggest that the proposed qCBCT approach has the potential to enhance tissue visualization when compared to clinically available CBCT during radiotherapy of prostate.