Dose warping uncertainties for the accumulated rectal wall dose in cervical cancer brachytherapy

Purpose: Structure-based deformable image registration (DIR) can be used to calculate accumulated dose volume histogram parameters for cervical cancer brachytherapy (BT). The purpose of this study is to investigate dose warping uncertainties for the accumulated dose to the 2 cm3 receiving the highest dose (D2cm3 ) in the rectal wall, using a physically realistic model (PRM) describing rectal wall deformation. Methods and materials: For 10 patients, treated with MRI-guided pulsed dose rate BT (two times 24 × 0.75 Gy, given in two applications BT1 and BT2), the planning images were registered with structure-based DIR. The resulting transformation vectors were used to accumulate the total rectum dose from BT. To investigate the dose warping uncertainty, a PRM describing rectal deformation was used. For point pairs on rectumBT1 and rectumBT2 that were at the same location according to the PRM, the dose for BT1 and BT2 was added (DPRM) and compared to the DIR-accumulated dose (DDIR) in the BT2 point. The remaining distance after DIR between corresponding point pairs, defined as the residual distance, was calculated. Results: For points within the D2cm3 volume, more than 75% was part of the D2cm3 volume according to both PRM and DIR. The absolute dose difference was <7.3 GyEQD2, and the median (95th percentile) of the residual distance was 8.7 (22) mm. Conclusions: DIR corresponded with the PRM for on average 75% of the D2cm3 volume. Local absolute dose differences and residual distances were large. Care should therefore be taken with DIR for dose-warping purposes in BT.