2020 AAMD Plan Study

Recap, Review, and High Performers

Recap

The 2020 AAMD Plan Study was a great success! The meeting was virtual this year, so the cloud-based ProKnow software fit perfectly well and allowed international participation. Over 300 people registered, and over 140 submitted valid, composite plans in the allotted time.

This year’s clinical planning and presentation team was led by dosimetrists Chavanon Apinorasethkul, CMD, and Kalin Shipman, CMD. They provided the anonymized planning images and anatomical structure set. They also led the design of the metrics and plan scorecard, which featured a total of 24 metrics studied (20 of which were scored; 4 others collected for added analysis) for a total of 150 points.

The featured plan was a base of skull chordoma. Participants were to design and submit a sequential plan prescribed to a total of 77.4 Gy and made up of two fraction groups: 54 Gy in 30 fractions followed by 23.4 Gy in 13 fractions. Participants used a dedicated ProKnow domain (https://aamd.proknow.com), where they could upload their plan(s) and dose(s), sum them to create a composite plan (if necessary), render their scorecard, and iterate until they were happy with their performance. Scores were kept blind (i.e., participants could only see their own plan and score) up until the AAMD presentation which took place on July 6, 2020. (Note: Visit the AAMD website to get access to their presentation.)

The scorecard’s metrics and objectives were based on current clinical goals for this type of treatment without extra challenges for target coverage objectives or ALARA scoring for organs-at-risk. As a result, an impressive number of participants met every highest objective over all metrics to achieve perfect scores of 150 out of 150.

A list of “High Performers” was defined as any/all users who scored in one of the “top three bins” in terms of overall plan score histogram (in this case, that means a score greater than 137.50 out of 150). Please see the “High Performers Summary” section, below.

Interactive Review

We used a dedicated ProKnow “big data” repository and analytics domain for this study rather than the conventional Plan Study quality system technology. This offers some nice advantages:

• Participants were able to sum their dose grids here to get their composite plan/score, when applicable.
• This domain allows retrospective analysis and interactive visualization of each and every anonymized submission.
• This domain also allows full navigation of population scorecards, histograms/variation per metric, and scatterplots/correlation for any pair of the derived or numeric custom metrics.

To study and analyze the results yourself, simply sign in to the domain, then use the many population analysis tools and of course the plan-by-plan analysis tools. The domain has been set to “read-only” mode, so you will not be able to edit contours, metrics, or scorecards, or upload any new data, but you can survey the plan submissions one-by-one or as a population.

Note: If you did not participate but would like to inspect and navigate the plans, doses, DVHs, scorecards, and population results, you can submit a request to Elekta/ProKnow team members at support@proknowsystems.com.

Figure 1. This is a histogram of the composite scores for the final valid, submitted plans. You can separate these by modality, TPS, or radiation type. You can also click on any bin to see all the plans whose metric fell in that bin. This goes for any derived or custom metric.

Figure 2. Here is a snapshot of the plan scorecard extracted across the whole population of submitted plans. The variation of each metric’s result is shown as a box plot atop the scoring objectives. Any row can be selected (double-clicked) for drill-down analysis of the underlying data.

Figure 3. Condensed boxplot view of all plan scores filtered by modality.

High Performers Summary

Congratulations to all participants, and especially to the following individuals who scored in one of the top performance bins of this plan study.

Table 1. The following users scored in the “top three bins” in terms of total plan score. For the cohort of users who scored 150, the order in the table is set simply by alphabetical order based on first name.