Session Item

Purpose: Toxicity from cervical brachytherapy has been demonstrated to mostly relate to the D2cc. Therefore a simplified version of knowledge based planning may be possible, only investigating the relationship of the overlap distance for 2cc and the D2cc from planning. This study aims to investigate if such a simple relationship could be used to predict the D2cc, and if this can be used to detect sub-optimal plans and improve plan quality.

Methods: The overlap volume histogram (OVH) method was used to determine the distance for 2cc of overlap between the OAR and CTVHR. Linear plots were used to model the OAR D2cc and OVH predicted 2cc distance. The model was used to predict D2cc, and this value entered as the maximum constraint in the inverse planning algorithm. Two datasets of 20 patients (plans from 43 insertions in each dataset) each were used to create 2 models, and the performance of each model was compared. Doses were scaled to ensure consistent CTVHR D90 doses.

Results: Mean bladder doses decreased by 2.9% for both models, mean rectal dose decreased 14.9% for model 1 and 6.0% for model 2, mean sigmoid dose decreased 10.7% for model 1 and 6.1% for model 2, mean bowel dose decreased 4.1% for model but no statistically significant difference observed for model 2.

Conclusion: A simplified knowledge based planning method was used to predict D2cc and was able to automate optimization of brachytherapy plans for locally advanced cervical cancer.