ESTRO 2020 Abstract book

S876 ESTRO 2020

Conclusion: Radiomic features from dose maps seem to add a significant value to usual NTCP models and could help better assess the risk of clinically relevant radiation induced lung toxicity. Evaluation of the trained models in another cohort is currently undergoing. PO-1531 Publishing linked and FAIR radiomics data in radiation oncology via ontologies and Semantic Web A. Traverso 1 , M. Vallieres 2 , J. Van Soest 1 , L. Wee 1 , O. Morin 3 , A. Dekker 1 1 Maastricht Radiation Oncology MAASTRO clinic, Radiotherapy, Maastricht, The Netherlands ; 2 McGill University, Medical Physics, Montreal, Canada ; 3 University California San Francisco, Radiation Oncology, San Francisco, USA Purpose or Objective Radiomics showed promising results in radiation oncology. However, most of the radiomics studies remain difficult to reproduce because of poor quality of reporting and the adoption of the so-called black-box approach. This lack of transparency is limiting the generalizability and validity of radiomics-based models. In this study, we present a proof- of-concept study using our newly developed radiomics ontology (RO), combined with Semantic Web technologies, as instrument for enabling interoperability of radiomics data following FAIR (Findable Accessible Interoperable We developed the radiomics ontology (RO, available on BioPortal): 458 classes and 76 predicates covering the whole spectrum of the workflow of radiomics computation, as presented in the IBSI (Image Biomarker Standardization Initiative). As a proof of concept, two institutions used two different open source radiomics packages to extract features from gross tumor volumes scans from publicly available CT scans of lung cancer patients. Features and metadata were converted from relational databases to RDF (Resource Description Framework) triples and uploaded to a SPARQL endpoint (Figure1a), as per FAIR guidelines. Reusable) principles. Material and Methods

Conclusion In a cohort of grade II/III glioma patients, symptomatic or clinically silent contrast-enhancing lesions following proton therapy clustered in direct proximity to the ventricles and around the CTV boundary. The data indicate substantially increased cerebral radiosensitivity of the PVR as well as a variable proton RBE and suggest the need for adapted treatment planning. PO-1530 Pulmonary toxicity in lung cancer treated by (chemo)-radiotherapy : a radiomics-based NTCP. V. Bourbonne 1,2 , F. Lucia 2 , G. Dissaux 1,2 , B. Julien 2 , D. Visvikis 2 , O. Pradier 1,2 , M. Hatt 2 , U. Schick 1,2 1 CHRU Brest, Radiation Oncology, Brest, France ; 2 Univ Brest, LaTIM- UMR 1101- INSERM, Brest, France Purpose or Objective Purpose: (chemo) – radiotherapy is the gold standard therapeutic option for patients with locally advanced lung cancer non accessible or ineligible for surgery. Despite few advances in progression free survival and overall survival thanks to recents advances (i.e durvalumab), prediction of toxicities, namely lung toxicity and its inherent morbidity, remain insufficient. Current dose-volume histograms (DVH) do not account for spatial dose distribution and strict application of current dose constraints does not prevent serious toxicity for all treated patients. We aim to invest the added value of radiomics applied to dose maps to predict acute and late pulmonary toxicity. Material and Methods Methods: Acute and late toxicities scored using the CTCAE v4.0 were retrospectively collected on all patients treated by arctherapy-based (chemo)-radiotherapy in our institution. Radiomic features were extracted from 3D dose maps from homolateral, controlateral and both lungs. After selection of the features based on a ROC (receiver operating characteristic) analysis, dose distributions, clinical factors (age, performance status, AJCC stage, baseline respiratory function, histology type,…) and radiomic features were logistically combined in order to train three models a clinical, a clinical + DVH and a clinical + DVH + radiomics based models. These Normal-Tissue Complications Probability (NTCP) models were evaluated for prediction of clinically relevant pulmonary toxicity: acute pulmonary toxicity (APT), late pulmonary toxicity (LPT), radiation induced pneumonitis (RIP) treated by corticosteroids (RIPCTC) and RIP needing hospitalization (RIPHPT). Results Results: 167 patients were treated from 2015 to 2018: 38% squamous-cell carcinoma, 40% adenocarcinoma, 14% small cell, 8% other histology with a median age at treatment of 66 years. Respectively, 22.8%, 16.8%, 19.2%, 10.8% experienced an APT > grade 1, LPT > grade 1, an RIPCTC and an RIPHPT. Areas under the ROC curve (AUC) for APT, LPT, RIPCTC and RIPHPT were respectively of 0.63/0.65/0.82, 0.65/0.72/0.80, 0.62/0.69/0.82 and 0.63/0.79/0.85 when comparing the clinical, clinical + DVH and clinical + DVH + radiomics models. Radiomic features selected in the models were mostly those extracted from the ‘homolateral lung’ volume or the ‘both lungs’ volume. Conclusion

Results Each of the users could independently query all the features generated from the two Software, without having any prior knowledge of the original labels used to store features and associated computational details. Using SPARQL queries, we could reveal the parameters accounting for major differences in feature values as a function of mapped metadata, enabling interoperability between different radiomics computational packages. Finally, radiomics features were linked to corresponding clinical data as shown in Figure 1b. The group has released a python package of the whole proof of concept, which allows any radiomics user to adopt FAIR principles for its study.