The 2023 AAMD Plan Study - A Question of Time

May 3, 20235 minute read

Hi there, everyone! This short message is to announce the 2023 AAMD ProKnow Plan Study, which is, as always, open to anybody and everybody in the international radiation therapy community. The more people who participate, the more we all learn and move forward, so we hope that hundreds of you can carve out a little time in your schedule in the month of May to take part.

We are trying a unique strategy for this study — a strategy that introduces some different, but timely, learning objectives that have immediate relevance to our day-to-day craft.

Here, we will summarize these objectives and tell you the “next steps” for getting started.

A “Timely” Question

In conventional plan studies, you are provided a dataset with images and predefined anatomy, then given a challenging plan quality scorecard to use to create and assess your treatment plan. By looking at the population, and in particular, high-performing plans, we can learn which techniques, technologies, and human tacticians can score the highest. Then, we gather those methods and try to share them with you all, along with giving the interactive (and anonymized) population results and statistics for you to peruse. This has worked well for many studies and workshops over many years.

But one question this conventional approach fails to answer directly is this: “How long does it take to create those great plans?”

We live in a world of adaptive radiation therapy where an updated treatment plan is needed for a patient based on a new permutation of their anatomy. Maybe it’s between fractions (offline adaptive), or maybe the patient is waiting on the treatment table (online adaptive); in either case, time is of the essence.

In this study, we take direct aim at collecting data and building knowledge about how long it might take to build an acceptable adapted plan and how the efficiency might be predicted by planning technique, software models and versions, computing power, etc.

A New Twist to the Plan Study Methods

To do this plan study, we first put out a survey to the AAMD community to verify interest and help design our methods so that they would be both practical, useful, and trustworthy. That feedback helped a lot! The design is summarized below, and you can find it in greater detail in the instructions to the Phase 1 (of two phases) plan study that is published already.

The Scenario

You prepared an initial head/neck for a patient and had it evaluated and approved by the radiation oncologist using a comprehensive scorecard of metrics and objectives. During the course of treatment, the physician deemed that the patient’s anatomy had changed enough to warrant an updated plan. You are asked to create that plan *as soon as possible* and are given a similar scorecard (in terms of metrics) but simpler in terms of objectives (binary goals rather than multi-level).

The study question is as follows: How long does it take to create an acceptable updated plan if working uninterrupted and using the initial plan experience (e.g., starting with an existing established optimization template), however applicable?

The Steps to Participate

(1) Reserve a small chunk of time where you can work on the timed portion of the study conveniently and ideally without interruption. The important phase of the study will be open between May 5 and May 19, which gives you two weeks during which to find a couple hours of time.

(2) When convenient for you, do the Phase 1: Initial Plan part of the study, which will emulate the initial treatment plan and help you prepare your optimization template and knowledge for the subsequent adaptive plan.

  • Phase 1 will look like a conventional, scored plan study. However, there is NO need to spend inordinate time trying to optimize your score here. Rather, just create a practical, high-quality plan that you are happy with, following the basic instructions. The scores of the initial plans will be studied only in terms of their relationship to the ultimate efficiency metric of the adaptive plan and not as a primary focus.
  • Do this phase of the study whenever it’s convenient for you. You do not need to time yourself or reserve chunks of time in your schedule for this phase!

(3) On or after May 5th when the self-timed, Phase 2: Adaptive part of the plan has been opened, you can complete this — the critical part of the study!

  • Start your timer as soon as  you import the new images and structures into your planning system.
  • The scorecard for Phase 2 will have the same metrics as the initial plan (with perhaps a twist or two, emulating an ambitious radiation oncologist), but the objectives will be binary (either “goal met” or “goal not met”).
  • Your job is this: Using your initial plan as a starting point, you will design plan(s) for the new anatomy until you meet all the binary goals.
  • Once you have a plan that meets all the goals, stop your timer, then calculate your planning time (in units of hours, e.g., 1.25 hours).
  • Submit your adaptive plan to ProKnow and answer the survey questions, the most critical one of which is how long it took to create the adaptive plan that met all goals.

What We Hope to Learn and Share

The 2023 AAMD Plan Study team will collect all the results and do some useful number crunching and analyses to hopefully shine a light on the following questions:

  • What is the variation in planning time it takes to achieve all goals for all metrics for an adaptive plan?
  • What is the variation in adaptive planning time across different TPS models?
  • Is there any correlation in adaptive planning time vs. original plan score (i.e., does a higher quality initial plan makes adaptive planning faster)?
  •  Is there a correlation in adaptive planning time vs. user-specified software or hardware “grade?”
  • Does adaptive planning speed vary with years of treatment planning experience?
  • For any given TPS, what tips and techniques can speed up the process of designing an adaptive plan?

Thank you, and happy planning!