Room: ePoster Forums
Purpose: Due to stochastic nature of Eclipse VMAT optimization solver, sequential optimization runs with the identical optimization criteria may converge to different solutions. Thus repeated re-optimization may result in a better treatment plan for a patient. We investigate the variations of critical structure and target DVH parameters during repeated optimizations.
Methods: 3 types of patient plans were considered, prostate, prostate fossa, and prostate/fossa plus pelvic lymph nodes treatment, with 4 patients selected in each group. For each of the patients, a standard 2 arc 6MV planning template with a set of fixed optimization objectives was used to generate a treatment plan. Varian Eclipse 11 VMAT optimization module was run 6 times with the same input parameters for each patient. The resulted plans were normalized so that 95% of the PTV volume was covered by the prescription dose. For comparison purposes, a prescription of 7920cGy was used for all types of target plans. Variances of Bladder D50, D25, Rectum D50 D35 and D15, femoral heads D10, plan's hot spot and PTV D100 were calculated.
Results: Ranges and standard deviations for the monitored DVH parameters were calculated. The largest fluctuations observed were for femoral heads D10 parameter, where the range exceeded 2.9Gy for almost all patients, and reached 5.3Gy for a prostate+lymph nodes target patient. The range of hot spot values was about 1% for all patients. PTVmin dose varied in 2-3% range. For rectum parameters, D35 showed the largest variation, with 2Gy range for most of the patients, but with only 0.5Gy for 2 patients. Bladder parameters were the most stable, with 0.5-1Gy range for all the patients except one with bladder D25 range of 3.4Gy.
Conclusion: Repeated runs of the Eclipse VMAT optimization procedure may result in a dosimetrically more advantageous plan, with DVH parameters improvement of a few Greys.