ESTRO 2021 Abstract Book

S582

ESTRO 2021

(P MC /P DC /P AC /P ADC /P AAC , see Fig. 1). Dose differences were quantified using dose-volume histogram (DVH)

parameters: Heart D mean

, Heart D max

, Lungs-GTV D mean

, Lungs-GTV V 5

, Lung contra

V 5

, SpinalCord 03

D max

, MedEnv 05

D max . Inter-plan variations between treatment plans generated using clinically acceptable contours were quantified dosimetrically, and the influence of contour variation on treatment plan generation and evaluation was analyzed. , and Esophagus D max

Results Clinically acceptable treatment plans could be generated fully automated for almost all patients. For only 10/140 (7%) of the DVH parameters distributed over 6 patients, the DVH parameter was above the clinical constraint (3x Heart D mean , 1x Lungs-GTV D mean , 5x Lungs-GTV V 5 , and 1x SpinalCord 03 D max ). Inter-plan variability on clinically acceptable contours was highest for Heart D max (3.4±6.8Gy) and lowest for Lungs-GTV D mean (0.3±0.4Gy). Dose differences in plan evaluation between treatment plans (P DC /P AC ) on automatic contours (C DC /C AC ), and evaluated on the three clinical contours (C MC /C ADC /C AAC ) were highest for Heart D mean (C DC : 1.8Gy, C AC : 0.6Gy), Heart D max (C DC : 6.0Gy, C AC : 2.7Gy) and Esophagus D max (C DC : 8.7Gy, C AC : 2.3Gy). Average dose differences in plan generation using the various contouring methods, evaluated on C MC , were on average below 1Gy/1%, with 11% and 15% above 2Gy/2% for P DC and P AC respectively. See Fig. 2 for an example patient with corresponding DVH’s.

Conclusion Dose differences arising from treatment plans generated on automatic contour variations were of the same (or lower) magnitude as inter-plan variability. For the heart, it is recommended to always correct delineation errors because they result in larger differences in DVH values. For the lungs, mediastinum, and esophagus,