Radiation Oncology

Automation of Treatment Planning & Treatment Quality

Automated tools for patient safety are a research and clinical interest of UCLA. Ensuring the utmost levels of patient safety is part of the responsibility of the medical physicist in a radiation oncology clinic. Our research interest is to develop automated tools in order to improve the efficiency and reliability of our patient safety processes.

  • We have shown that x-ray setup images can be analyzed to detect errors in patient ID and setup by an automated algorithm.
  • 2D-3D registration by BrainLAB ExacTrac IGRT device was used as proof of concept. Any IGRT package using 2D-3D registration should perform equivalently (Varian, Elekta, and Accuray all have 2D-3D products).
  • Image similarity measure used for registration was used "out of the box" for error detection with 99% or better success rate. Probable improvement with purpose-build similarity measure.

Project: Assessment and prediction of treatment planning quality through integrated planning script

The goal of this project is to:

  • to develop quantitative metrics for plan quality evaluation and assessment
  • to improve intra- and inter-planner consistency
  • to identify plan quality prognostic features that can be used to improve treatment plan quality

The clinical plan workflow with integrated planning script is illustrated in Figure 1. An example of the plan report is shown in Figure 2 which is automatically generated by the planning script based on a self-evolving planning database and national published standards.

planning
Figure 1.

Figure 1. During treatment plan review, dosimetric quantities are automatically extracted by the planning script and compared against national published standards and institution-specific criteria generated from statistics of accumulated planning data. Upon acceptance of a plan, the plan quality indices and patient specific parameters are extracted by the script and saved back into the planning database to update the existing institution-specific planning statistics.

quality report
Figure 2.

Figure 2. A representative plan quality report consists of achieved dosimetric parameters (blue box), national published standards (red box) and statistics from institution plan data base (green box). Right: Graphic ranking of currently achieved parameters with respect to the statistics of accumulated planning data is shown in the bar plot as a part of plan quality report.