Structure Relations Presentation

[GregSal]

Presentation Content

1. Introduction
   1. What is the need
      • Good VMAT plan requires accurate and appropriate structures for optimization and evaluation
      • VMAT plan can have many derived structures. It important to verify that the have been created & labeled correctly.
      • Optimization strategies can also depend on Target-OAR overlap/proximity. Dosimetrists & physicists needs to have a clear picture about potential competing priorities.
   2. How dose this address that need
      • Structure Relations can generate a diagram and a table that display the relationships between target and OAR structures.
      • It will also provide numerical values for the amount of overlap between two structures and the distance between two structures.
      • Structure Relations diagrams can also display relationships between derived structures, such as margin expansions, optimization structures and avoidance structures
   • What it won't do
     • Structure Relations is intended as an aid for planning prep and QA purposes.
     • It will not modify contours.
     • It is not intended to evaluate the quality of the contouring
2. Example Workflow
   1. Preparing a planning strategy
      1. Dosimetrist receives planning CT and contoured structures from RO. It has 2 separate target volumes
      2. Structure set is sent to Structure Relations app which analyzes the contours.
      3. Dosimetrist generates a diagram showing all OARs that overlap with either of the two targets along with the percentage of overlap, and uses this to guide generation of OAR and target opt structures
      4. Dosimetrist generates a diagram showing the distance between the two targets and between the targets and select non-overlapping structures. They use this to guide the initial plan geometry and optimization criteria.
   2. Plan QA
      1. Physicist receives plan and contoured structures with 2 separate target volumes.
      2. Structure set is sent to Structure Relations app which analyzes the contours.
      3. Physicist generates a diagram showing the relationships between all target volumes an uses this to check:
         • That all structures are appropriately labeled e.g. CTV_1 contains GTV_1 and is within PTV_1
         • That appropriate expansion margins are used for the CTV and/or PTV.
         • That appropriate opt structures are present and created correctly: e.g. opt Bladder partitions Bladder and borders on PTV
         • When an expected relationship is violated, The physicists uses individual slice relationships and 2D contour comparisons to identify the cause of the violation. For example a GTV overlaps with its PTV, the physicists observes that for most slices the PTV contains the GTV, but only on two slices the GTV overlaps the PTV. Checking the CT for those slices they see that the GTV was based on a MIP CT, but the physician manually cropped the PTV back from the chest wall based on the average CT, causing it to overlap the GTV on those slices.
      4. The physicist generates a diagram that shows the OARs that overlap with either of the two targets along with the percentage of overlap, and uses this to focus their attention on aspects of the plan where target / OAR tradeoffs are most critical.
3. Example Output
   • show some examples of current diagrams generated for plan checking
   • Relationship matrix
   • Slice based contour comparison
4. Algorithm design
   • Slice based relations and metrics with 3D corrections
   • Relationship definitions
   • Tree diagram of relationship tests
5. Next Steps
   1. Add numeric (margin/distance/volume ratio) display options to diagram and matrix
   2. Improved diagram layout design
   3. Add CT images to slice-based display
   4. Add site-based testing templates with expected relationships, and margins that generates a report highlighting failed tests
   5. Add ability to annotate diagrams illustrating problem relationships or metrics