ESTRO 2023 - Abstract Book

S498

Sunday 14 May 2023

ESTRO 2023

Conclusion Practice has changed in our department over the last 8 years, with fewer patients being offered long-term palliative RT. With the realization, that 39Gy/13F offer no benefit in OS, this treatment regime is no longer offered at our institution. The change towards shorter RT courses is likely in part due to the COVID pandemic, which presented a demand for short course RT with few appointments for each patient. We do however believe that the increased use of short course RT based on improved patient selection, is to the benefit of elderly and frail patients, who will spend less time in the hospital and may suffer less toxicity compared to long course RT.

Proffered Papers: Dose accumulation and dose prediction

OC-0613 Automatic treatment selection guided by deep learning: a proof-of-concept for esophageal cancer C. Draguet 1,2 1 UCLouvain, IREC/MIRO, Brussles, Belgium; 2 KULeuven, Laboratory of Experimental Radiotherapy, Leuven, Belgium Purpose or Objective The model-based approach is clinically used in the Netherlands to refer a patient to conventional radiotherapy (XT) or to a proton therapy (PT) treatment. In this context, the benefit of using deep learning (DL) models to predict XT and PT plans are substantial. The plan generation becomes automatic and the planners’ various expertise has no longer an impact. In this study, we develop an automated model-based approach by combining the use of such DL dose prediction models for intensity-modulated radiation therapy (IMRT) and pencil beam scanning (PBS) treatment with a normal tissue complication probability (NTCP) model. The accuracy of this clinical decision tool will be evaluated. Materials and Methods Two U-Net architectures with dense connections were trained: one for XT plans and one for PT plans. They were trained on a database of 40 esophageal cancer patients using cross-validation and a circulating test set to predict a dose distribution for each patient. The U-Net models take the CT scan, the contours of OARS and of the target volume as inputs. We also investigated an alternative approach where a first estimation of the dose distribution (D_0) is passed along with the other inputs in the DL models. D_0 is generated on the RayStation software by running a few optimization steps on a template of OARs and target objective functions. No manual fine-tuning is needed. The DL models were chained with a NTCP model for postoperative pulmonary complications. The model parameters are the age, the histology, the BMI and the mean lung dose (MLD). The accuracy of our DL models for patient referral will be evaluated by using ∆ NTCP between XT and PT plans. The protocol from the Dutch Society for Radiotherapy and Oncology recommends the use of a model-based approach with a ∆ NTCP threshold value ≥ 10% to refer a patient to PT. Results Our DL models succeed in predicting the dose distributions. Dice coefficients between the prediction and the manually generated plan (groundtruth) range from [0.844,0.946] for the XT model without prior-knowledge (standard model) and [0.902,0.98] for XT with prior-knowledge. For PT, the standard model reaches Dice values in the range [0.917,0.956] while it is in the range [0.945,0.981] for the prior-knowledge model. Figure 1 shows the clinical decision that is made while chaining our DL networks with the selected NTCP model. All patients are correctly classified with both approaches.