Room: Exhibit Hall | Forum 4
Purpose: Investigate the use of contour propagation to streamline HDR tandem and ring volumetric planning and support single MR-scan and per-fraction CT scans.
Methods: Ten cervical cancer patients treated with HDR ring and tandem post external beam (EB) were simulated on CT and MRI for each fraction. A single physician contoured the high risk (HR)-CTV and organs at risk (OAR) on the CT dataset for each fraction. The MRI was utilized to assist in contouring the HR-CTV. The contours were propagated from fraction 1 to subsequent fractions using rigid registration (RR) and deformable image registration (DIR). Plans were inversely optimized with the intent to meet the EMBRACE I objectives. Five plans were created for each fraction: (A)all contours manually drawn, (B)OARs manually contoured and the HR-CTV propagated using RR or (C)DIR, (D) all contours propagated using RR or (E)DIR. The Dice Similarity Coefficient (DSC) was calculated for the contours using RR or DIR. Dose metrics were compared between plans to assess the accuracy of the propagated contours. An analysis was performed to determine whether the difference in average dose metrics for each plan type was statistically different from plan 1.
Results: On average, acceptable volume-based plans were achieved that met OAR constraints with the exception of the bladder for plans D and E, where all contours were propagated from fraction 1. Target coverage was compromised to maintain OAR constraints, and was similar between plan types. No statistical dose difference was observed, except for the bladder, between plans. The DSC results were comparable for the HR-CTV between the RR and DIR techniques and between the two DIR software packages.
Conclusion: Dosimetrically equivalent plans were achieved regardless of whether the HR-CTVs were manually drawn or propagated for plans A-C, suggesting post EB, a single HR-CTV contour may be propagated across fractions.