Room: Exhibit Hall | Forum 2
Purpose: Adaptive radiotherapy (ART) can often achieve better delivered dosimetry when there is considerable change of patient anatomy. One challenge is the quick estimation of the dose distribution on the actual anatomy of treatment day. The project aims to evaluate the accuracy of registration-based dose transfer on treatment CT.
Methods: Five previously treated prostate cancer patients were included in the study. A 2-arc VMAT plan and a 9-field IMRT plan were generated for each patient with 180 cGyx25 fractions (4500cGy in total) with daily CT-on-rails image guidance. For each patient, 5 weekly treatments with noticeable anatomic change were selected to perform dose estimation. Contours on the planning CT were transferred to treatment CTs through registration then manually adjusted. For each selected treatment, beams from the original plan were directly calculated on the daily treatment CTs and dose was used as reference (DC dose). In addition, doses were both rigidly and deformably transferred from planning CT to treatment CTs respectively (RT dose, DT dose) using image registration. The fractional dose distribution on target and organs-at-risk (OAR), rectum and bladder, were compared using a common clinical goal.
Results: For target, significant difference (p<0.01) occurred between recalculated dose and rigid transferred dose. In VMAT plans, average difference of V100% on 4500cGy prescribed dose was 16.49%Â±13.32% for PTV. For rectum, average V40Gy difference was 6.63%Â±5.84%. For bladder, average D10cc difference was 0.93%Â±0.69%. Similar data can be observed from IMRT plans. Deformable transfer method showed much larger dose difference compared with the recalculated dose.
Conclusion: With common dosimetric protocol used for the prostate cancer radiotherapy, we found dose transfer via rigid fusion method appeared to be inaccurate for the target coverage and rectum hot spot, but reasonably accurate for dose estimation on bladder. Deformable dose transfer shouldn't be applied to dose estimation.