Program Information
Knowledge-Based Automatic Treatment Planning for Prostate IMRT Using 3-Dimensional Dose Prediction and Threshold-Based Optimization
M Folkerts1,2*, T Long2 , R Radke3 , W Lu2 , X Jia2 , X Gu2 , M Chen2 , S Jiang2 , (1) University of California, San Diego, La Jolla, CA, (2) UT Southwestern Medical Center, Dallas, TX, (3) Rensselaer Polytechnic Institute, Troy, NY
Presentations
SU-E-FS2-6 (Sunday, July 30, 2017) 1:00 PM - 1:55 PM Room: Four Seasons 2
Purpose: Demonstrate the feasibility and performance of a knowledge-based automated planning (KBAP) algorithm consisting of 3-dimensional (3D) dose distribution prediction and threshold-based fluence map optimization for prostate IMRT plans.
Methods: We developed a 3D dose prediction model for iso-dose contours from 2% to 100% of Rx Dose (at 2% intervals) using PTV and OAR geometric information as input. From a library of 65 clinical 7-beam prostate IMRT plans, we performed leave-one-out (LOO) tests, with the prediction model being "trained" using the remaining plans. For each test, we performed a fluence map optimization using predicted dose as threshold values for each voxel. The optimization objective function penalizes deviations from threshold values in the following situations: overdose for OARs and PTV, and under-dose for PTV. Additional terms penalize PTV deviation from Rx Dose and voxels above mean dose in each OAR. Final doses for both the original clinical plan and the KBAP plan were calculated with Monte Carlo. We then compared the resultant plans both qualitatively and quantitatively.
Results: We observed PTV coverage in clinical and KBAP plans to be practically identical. When comparing KBAP plans to clinical plans, the average OAR sparing between D15% and D50% ranged from 5.9 to 8.3 percent of Rx dose for bladder and 0.8% to 1.5% of Rx dose for the rectum. The average reduction in OAR volumes receiving greater than 50% of Rx dose ranged from .3% to 5% for bladder and -1% to 1% for the rectum. We observed several cases where OAR trade-off doesn't match the original plan; for example, more dose to the bladder and less dose to the rectum (or vice versa).
Conclusion: KBAP plans achieve clinical quality PTV coverage and similar if not better OAR sparing for the prostate IMRT plans we tested.
Contact Email: