MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Knowledge-Based RapidPlan Models for Left and Right Breast Using RapidArc

O Apaza*, A Garcia, M Almada, C Venencia, Instituto Zunino - Fundacion Marie Curie, Cordoba, ARGENTINA

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose:
The objective of this work was to create and validate RapidPlan models for breast using RapidArc


Methods:
50 clinically breast RapidArc plans were selected to create two RapidPlan models (breast_left and breast_right). The plans were generated in Eclipse v15.5 (Varian) with 6MV of a Novalis Tx (Varian-BrainLab) equipped with a high resolution multileaf. The models were evaluated on the basis of the statistical goodness of fit using the Pearson correlation coefficients, average chi squared and the goodness of estimation through the mean square error. For validation: 20 plans that integrate the models were reoptimized with RapidPlan (closed validation). 20 plans that do not integrate the models were optimized with RapidPlan (open validation). Dosimetric parameters of interest were compared between clinically accepted plans and RapidPlan (closed and open validation): contralateral breast dose (Mean dose / Dmax), heart (D8% / Mean dose) and homolateral lung (D50% / D20% / D10%).

Results:
The statistical goodness of fit and the estimation for the structures of interest (heart / homolateral lung) in both models were for right breast (R2 =0.47, ?2 =1.094, MSE =0.05 / R2 =0.41, ?2 =1.058, MSE =0.05) and for left breast (R2 =0.602, ?2 =1.026, MSE =0.01 / R2 =0.41, ?2 =1.084, MSE= 0.05). For closed validation: breast_right neither contralateral breast (p = 0.410 / p = 0.578) nor heart (p = 0.718 / p = 0.907) did not show statistically significant differences unlike right lung (all p = 0.001) and breast_left neither contralateral breast (p = 0.141 / p = 0.163), nor heart (p = 0.155 / p = 0.115) nor left lung (all p > 0.148) did not shown statistically significant differences. For open validation in both models no statistically significant differences were obtained.

Conclusion:
Two RapidPlan models for VMAT simple breast were successfully implemented.

Keywords

Breast, Treatment Planning

Taxonomy

TH- External Beam- Photons: Treatment planning using machine learning/Knowledge Based Planning/automation

Contact Email