Plan Practicality Guidelines, Additional Info, and More FAQs
The 2019 World Championship event 24-hour participation window opens at 6:00 a.m. (Eastern Daylight Time, or UTC-04) on Monday, September 9, 2019.
Below are some final instructions, including a description of a few general “plan practicality” guidelines. We also include a few more FAQs to add to previous updates.
This is the final update before the event. Thank you for participating, and good luck to all of you. And don’t forget — have fun!
Plan Practicality Guidelines
One crucial and ever-present rule applicable to all ProKnow plan studies should be followed: always submit a final plan that your clinic would accept for actual patient care in terms of plan practicality, dose algorithm used, or other planning methods. If you are manipulating your plan to increase your score but make it less feasible or efficient, then you are straying from the intent of these studies. If you are to compete, compete fairly!
Dose Calculation Grid Size
You must use a dose calculation grid size of 3.0 mm or less in each dimension. That said, use what you would clinically use for this type of patient, so if 2.0 or 2.5 mm is normal, then you should use that.
Specific Guidelines for Photon (or Photon/Electron) Plans
In addition to that general rule, we are providing a few specific guidelines regarding external beam plans designed with photons and/or electrons (e.g., IMRT, VMAT, helical, robotic).
- If you are using VMAT, you may use up to four (4) fields, or full arcs. If your TPS automatically splits full arcs into multiple RT beams (e.g., if to achieve avoidance regions you must use multiple sub-arcs), those component arcs will not count as separate fields.
- If you are using IMRT, you may use up to nine (9) fields, but not more.
- Total monitor units will be monitored carefully, so avoid overly complex modulation that you would not use clinically.
- The field of view of the planning CT crops some portion of the shoulders. As you would for a clinical plan, you must obviously take care to avoid portions of modulated beams contributing entry dose through those regions of the shoulders. You will notice that the scorecard will have two metrics/objectives that will be punitive if there is too much dose (i.e., significant beam entry dose) in those avoidance regions. There will, of course, be some dose there due to scatter and exit dose from beams entering other regions that are not cropped.
- Use photon energies of 10 MV or less (e.g., 6 MV, 6FFF, 10, 10FFF).
- While non-coplanar beams are not expected for a head and neck case such as this, they may be allowed if they are deemed practical and beneficial upon review.
There is an ongoing debate about the fairness of lumping proton plans in with IMRT, VMAT, and other external beam plans. This puzzle, in truth, is not yet solved. Interestingly, this debate stems primarily from proton planners who complain that some of the highest proton plan scores are probably not plans that would be safely or practically treated in the real world. For example, there is also the lack of modeling of “plan robustness” in ProKnow plan studies as they are today, and we all know how critical this is when designing and accepting proton plans in the real world.
Proton/ion plans will, of course, be accepted for this competition. However, because of this debate (and truthfully, because of our own lack of experience in judging the practicality of proton/ion plans) we have decided that proton/ion plans will not be considered for the top three cash prizes. We apologize if this seems discriminatory, it is not. Rather, the organization donating the cash prizes (RTU) wants to avoid the controversy for their cash prizes.
Proton plans will be included in the comprehensive results and big data analytics that will be published for the entire population of plans across all TPS and treatment modalities. It is expected that proton/ion plans will be self-audited by the expert planners with experience in determining the practicality of these special plans.
Top scoring plans will be audited, especially those eligible for prize money. Be advised that one form of audit will be to do dose QA via independent recalculations with high-quality 3rd party dose calculators and standard beam models for the users’ machine models (results will be normalized and compared to users’ dose grids). This will help verify that (1) the participants do not use any older and unacceptable dose algorithms (e.g., pencil beam photon) and (2) density heterogeneities are factored into the dose calculation. Offending plans will be disqualified.
Also, remember that each submission will be anonymized and included in the library of submitted plans for review by all participants after the event is over. Be sure to self-audit or else your peers may notice questionable plans or results!
Scorecard Rule: You Must Avoid “Unacceptable” Metric Results
After one of our public plan studies last year (TROG 2018), we got a few complaints that some high coring planners would willingly sacrifice a low-weighted metric (i.e., fail, and get zero points) in order to maximize points for a higher-weighted metric. While this may make sense for “optional” objectives, it is not good for clinically-required metrics.
For this competition, please treat every metric as clinically-required and do not allow any of your results to fall in the “UNACCEPTABLE” performance bin. That is, for your final plan, please ensure that none of the scorecard rows show up as UNACCEPTABLE (red color). Any unacceptable results on the scorecard will disqualify you from the overall and sub-category awards.
As explained in prior updates, you will be given a partial (i.e., incomplete) DICOM RT Structure Set (RTSS) along with the planning CT images. You will be able to quickly identify (by comparing the scorecard metrics with the partial RTSS) which structures you will need to contour, and how they should be named.
There will be a total of six (6) missing structures that you will contour (not counting additional structures you might add to aid in optimization, etc.). The gold contours for each of these six structures, which you will not see until after the results are published, have been reviewed and edited by a team of inter-institutional physician experts. Each of the six is important because they are each represented in the score. Two of them, however, will carry additional importance because their DICE coefficients (your contours vs. gold) will be part of the composite score. We will not tell you which two of the six will be analyzed this way until after the event.
We will reveal one of the six missing structures to you now: the larynx. We tell you this now so that you can know in advance that we want you to call the structure the “Larynx” but to include the glottic and supra-glottic volumes in this structure. (Note: This is also a great way to reward those of you who actually read these instructions carefully, as attention to detail is a vital skill in radiation treatment planning. If you’re reading this now, congratulations, you’re already performing at a high level!)
All of the other structures (e.g., body, targets, other organs-at-risk) will be provided to you in the partial RTSS. This large group of structures has not been comprehensively reviewed by the inter-institutional physician team, but that is less important for these because each participant will be using identical data for these structures. In some cases, you may see differences from your practice (e.g., CTV or PTV boundaries) but in any case, do not edit these structures! Use them just as they are; they are control variables in this particular study.
Contrast in Planning CT Images
You will notice that the CT images have contrast. While some centers may do special contouring of high-contrast regions (e.g., vessels) in order to do bulk electron density override of those regions for dose accuracy purposes, you do not need to do this for this study.
You must use heterogeneity corrections in your dose calculation, but you do not need to factor out the small volumes of contrast.
Make sure you take a moment to review all of the published FAQs from this prior update. We have added a few to that article since the first notification.
In addition to those, we would like to add two more FAQs that came up during Practice Week.
Upload Files Directly from within Patient (and Avoid Deleting Your Patient!)
When you upload your data (structure set, plan, and dose), be sure to be “inside” your assigned patient module (i.e., looking at your dataset) and then use the Actions → Upload File(s) tool. Do not use the Uploads button on the left of the screen.
Also, we have noticed that a few participants in the practice phase have deleted their patient, thinking they can re-upload their image set, structure set, plan, and dose from the main Uploads module. This results in an Incomplete submission and the message “Patient not found” on the competition dashboard. Please make sure you are uploading your structure set, plan, and dose submissions directly to your assigned patient. If you need a refresher on how to do this, please check out the walkthrough on our previous post: Practice Week! Preparing for the Championship Event.
If you accidentally delete your patient during the event, we can help restore it for you. However, this is a manual process and will take time. As such, it will affect your end time and potentially the efficiency component of your score.
Activating Patient Objects & Updating the Patient Hierarchy
A couple of users have had issues on the Patient Browse tab. If you’re having difficulties, here are some tips to get you “unstuck.”
To update the patient hierarchy to properly associate your uploaded files, you’ll first want to press the Edit button. You’ll notice the interface changes slightly. Identify the plan you wish to move, and hover over the icon with the three horizontal lines. Click and drag the plan and dose under the structure set. You’ll notice that the structure set icon will change to a green checkmark when you’ve moved it to a valid position. Release to drop it into place.
To activate an object in the patient hierarchy, you’ll need to double-click the object. Double-clicking an object will activate it and all of the objects up the hierarchy. For example, double-clicking the dose in a properly associated patient hierarchy will activate the dose, plan, structure set, and image set.