Click here to


Are you sure ?

Yes, do it No, cancel

Knowledge-Based Planning for Whole Breast Irradiation Using Tangential Beams

A Fernandez Toledo*, M Huynh-Le , X Trent Ning , J Hoisak , P Hartman , C Yashar , K Moore , L Cervino , UC San Diego, La Jolla, CA


(Sunday, 7/29/2018) 3:30 PM - 4:00 PM

Room: Exhibit Hall | Forum 4

Purpose: Because standard knowledge-based planning (KBP) methods are designed to operate in the context of inverse-optimized radiotherapy, standard-of-care forward-planned whole breast radiotherapy without a target contour presents a challenge for automated KBP. The purpose of this work is to construct a KBP dose prediction model and automatic planning workflow for whole breast irradiation using tangential beams. 

Methods: We have constructed DVH estimation models of the right and left breast using 73 and 62 previously treated plans, respectively, in RapidPlan™ (Varian Medical Systems). An in-house Matlab-Java code creates the structures needed for knowledge-based DVH estimation and automated planning: irradiated volume (defined by field projections), breast volume (using irradiated volume and radial expansion of lungs and heart), and skin flash structures. Using patient-specific DVH estimations to guide static field inverse-planning optimization objectives, fully-deliverable KBP-driven automated plans for 20 patients were created and compared against the clinical plans. DVH parameter differences δD=D(Clinical)–D(KBP) (Dmean and Dmax) and δV=V(Clinical)−V(KBP) (V10,V20,...,V90) for breast PTV and OARs were used to compare plans. In addition, two medical physicists and two radiation oncologists performed a comprehensive comparative study.

Results: The standard deviation of δV was 0.4% (Lung RT), 1.3% (Lung LT) and 1.2% (Heart), implying strong concordance between estimated DVHs and clinical DVHs. When converted into deliverable breast treatments, KBP plans on average demonstrated better histograms, deviations, and mean values for deltas for both target and OARs. In the comparative study, 75% of KBP plans were deemed clinically acceptable, and the other 25% would only need minor adjustments to be satisfactory.  31% of the KBP plans were considered better than or equivalent to the clinical plans.  

Conclusion: We have developed a KBP workflow for tangential-beam whole breast irradiation. Overall, the fully-deliverable KBP plans achieve similar quality as the previously-treated clinical plans included in the model.


Breast, Inverse Planning, Segmentation


TH- External beam- photons: IMRT dose optimization algorithms

Contact Email