Implementation of an Automation Tool for Treatment Planning Constraint Designation and Plan Evaluation

Applying dose constraints for individual structures and assuring that constraints are met represent critical components of radiotherapy treatment planning, particularly within the inverse planning process. Structure constraints defined for a variety of treatment site-specific target and normal tissue structures have previously been evaluated at our center via manual measurement from calculated dose statistics within the treatment planning system. This process is time consuming and prone to manual errors. A novel commercial system was implemented to automate this process. Eight treatment planners retrospectively defined and evaluated plan constraints for a total of 64 patients representing 10 distinct organ sites. The manual and automated processes were timed to evaluate efficiency. The accuracy of this system was then evaluated through the assessment of 862 structure constraints for these 64 patients. No deviations were observed between dose metrics evaluated manually and by the automated system. However, a total of 30/862 (3.5%) of the manually reported constraints differed by more than 1% from results from the automated system. While most differences were small and due to gross estimation of the constraint, some were clinically relevant mistakes, with 7 (0.8%) of them greater than 10%. Mean time for definition, evaluation, and documentation of these constraints was 6.2 and 1.9 minutes for manual and automated processes, respectively. The manual process was more time consuming for all treatment sites with minimum and maximum differences for individual treatment sites of 1.6 and 6.3 minutes, respectively. Due to increased evaluation simplicity and efficiency, the automated process has led to a >30% increase in the number of constraints evaluated per plan, thus providing a more complete dosimetric analysis of the target and organs-at-risk. In addition, this system facilitates near real-time plan metric evaluation and iteration, since structure metrics can be displayed within seconds for each iteration of the plan. It also allows simple evaluation of complex metrics without additional planning requirements, such as the creation of structure contours from isodose surfaces to evaluate common plan quality indices. Automated structure constraint definition, evaluation, and documentation results in greater accuracy and safety, reducing the rate of clinically relevant errors (>10% deviation) in plan metric reporting from 0.8% to 0%. In addition, the ability to evaluate all planning goals simultaneously during plan creation results in more efficient plan optimization and realization of planning goals. Automation of this process facilitated an increase in number of constraints evaluated while saving an average of 4.3 minutes per plan optimization.