Vienna, Austria

ESTRO 2023

Session Item

Poster (Digital)
KB multi-institutional plan prediction of the left whole breast irradiation with tangential fields
Alessia Tudda, Italy


KB multi-institutional plan prediction of the left whole breast irradiation with tangential fields

Roberta Castriconi1, Alessia Tudda1,2, Giovanna Benecchi3, Elisabetta Cagni4, Francesca Dusi5, Anna Ianiro6, Valeria Landoni6, Tiziana Malatesta7, Aldo Mazzilli3, Guenda Meffe8, Caterina Oliviero9, Lorenzo Placidi8, Giulia Rambaldi Guidasci7, Alessandro Scaggion5, Valeria Trojani4, Claudio Fiorino10

1IRCCS San Raffaele Scientific Institute, Medical Physics Department, Mian, Italy; 2Università Statale di Milano, Physics Department, Milan, Italy; 3University Hospital of Parma AOUP, Medical Physics Department, Parma, Italy; 4Department of Advanced Technology, Azienda USL-IRCCS di Reggio Emilia, Medical Physics Unit, Reggio Emilia, Italy; 5Veneto Institute of Oncology IOV–IRCCS, Medical Physics Department, Padua, Italy; 6IRCCS Istituto Nazionale dei Tumori Regina Elena, Medical Physics, Rome, Italy; 7Fatebenefratelli Isola Tiberina - Gemelli Isola, UOC di Radioterapia Oncologica, Rome, Italy; 8Fondazione Policlinico Universitario Agostino Gemelli IRCCS, UOSD Medical Physics and Radioprotection, Rome, Italy; 9University Hospital, “Federico II”, Medical Physics , Naplse, Italy; 10IRCCS San Raffaele Scientific Institute, Medical Physics Department, Milan, Italy

Show Affiliations
Purpose or Objective

Whole-breast irradiation with tangential fields (TF) is still a largely used technique in post-operative radiotherapy of breast cancer. The expected consistency in contouring and plan techniques together with the high number of treated patients may translate into knowledge-based (KB) models usable on a large scale. In the context of a multi-institutional project, we already demonstrated the transferability of KB-models for the right-breast TF irradiation. Aims of the present study were to evaluate inter-institute KB-model’s variability and transferability also in the case of the left-breast irradiation.

Material and Methods

Eight institutions set KB models by using RapidPlan (Varian Inc.), following previously shared methodologies. The plan prediction performances of the resulting models were tested on 16 randomly chosen additional patients (pts), available at the time of the investigation (2 pts per centers). All models were run considering the most used scheme, delivering 40Gy/15fr. DVH prediction bands of OARs (heart, ipsilateral lung, contralateral lung and contralateral breast) were analyzed. The inter-institute variability was quantified by the inter-institute SD of predicted DVHs/Dmean. The transferability of models among institutes, for the heart and the ipsilateral lung, was evaluated by the overlap of the geometric Principal Component (PC1) when applied to the test patients of the other 7 institutes. This parameter was previously assessed as robustly representative of the association between anatomy/segmentation and DVH prediction of OARs, in quantifying if a model can be reasonably applied to patients scanned and segmented in other Institutions.


Figure 1 shows the mean DVH of the ipsilateral-lung and heart (over all 16 patients) within the SD variability. Inter-institute SD of the DVH prediction in the dose range from 20% to 80% was 1.7% and 1.2% for the ipsilateral lung and the heart, respectively. The inter-institute variability in terms of Dmean was 0.5 Gy and 0.3 Gy for the ipsilateral lung and the heart, respectively; while for contralateral OARs the SD was <0.2 Gy. Figure 2 shows the distribution of the mean predicted heart and ipsilateral lung dose for every institute through all patients of the test set. Average institutional values were between 1.1 and 2.3 Gy for the heart and between 4.3 and 5.9 Gy for the ipsilateral lung. Models showed high transferability in terms of PC1 for both in-field OARs: the value of PC1 for each model in the test set was within 90th percentile of the corresponding training set in more than 80% of cases (80-100%), suggesting no relevant differences among models in terms of transferability.


Results show limited inter-institute variability of plan prediction KB models, and high model’s transferability. These findings encourage the building of a robust benchmark model, with large potentials for plan QA, audit, education/tutoring and large-scale KB plan automation.
This study is supported by an AIRC grant (IG23150).