Program Information
Automated Treatment Planning: Generating Choices Using Population-Based DVH Tradeoff Exploration
M Folkerts1,2*, W Lu2 , S Jiang2 , T Long2 , (1) University of California, San Diego, La Jolla, CA (2) University of Texas Southwestern Medical Center, Dallas, TX
Presentations
SU-E-FS2-2 (Sunday, July 30, 2017) 1:00 PM - 1:55 PM Room: Four Seasons 2
Purpose: Automatically generated treatment plans can offer improved efficiency in the clinic. However, plan quality and customization remain a challenge. An auto-generated plan will often require additional manual interaction to improve OAR sparing and explore possible trade-offs to satisfy physician curiosity. This work develops and evaluates an auto-planning tool which generates a set of plans consistent with major historical trade-off preferences found in previously approved DVH data from a population of treated patients. This tool gives the physician the ability to make an informed choice on which plan is best for their patient.
Methods: Principal component analysis (PCA) was used to find major modes of variations in OAR DVH in a database of 59 treatment plans. The major modes, along with the average PTV and body DVH, are used to generate a set of target DVH spanning historical trends and trade-offs between the bladder and rectum. We crafted a set of four target DVHs representing: the population average, overall OAR sparing, sparing bladder/sacrificing rectum, and sparing rectum/sacrificing bladder. These target DVHs are used in a threshold-driven optimization for reference-based auto-planning (TORA) to generate four treatment plans. We evaluated this trade-off exploration method on 40 patients and report on qualitative observations.
Results: The trade-off exploration produced a wide variety of plan choices for each patient case with D15% and D50% for both bladder and rectum varying by up to 20 Gy. We also observed variation in PTV coverage based on the OAR trade-off and sparing attempted. With PTV dose normalized to D95% = 81 Gy, we observed PTV D50%s varying between 2 and 8 Gy.
Conclusion: The results of this study show promise in providing physicians with choices from an automated planning system. Our results are primed for blind physician review, which we plan to pursue in the near future.
Contact Email: