ESTRO 2023 - Abstract Book

S1891

Digital Posters

ESTRO 2023

Materials and Methods The PREDMORN study was designed as a retrospective case-control study involving five European and one US teaching hospitals. The study hypothesises that the risk of developing ORN can be accurately predicted on an individual level with NTCP models based on dosimetric parameters and clinical and demographic variables. The participating centres joined the study at different time points over a period of 11 months. The data transfer process was initiated in parallel to protocol development (Humbert-Vidan et al. 2022) and local data collection processes. Results The use of each collaborating centre’s data is dependent on obtaining appropriate local governance/ethics approval and data transfer agreements (DTA). We initially explored a ‘Joint Controllers’ data transfer approach in order to facilitate a joint use of the data by all parties for the study. However, from a legal perspective, a ‘Controller to Controller’ approach was considered more efficient. For expediency, it was decided to proceed with individual DTAs between each participating centre and the receiving centre, seeking permission for the receiving centre to use the data for the study only. Data received under these DTA’s cannot be shared with other participating centres for others’ own ORN studies, thus limiting collaborative efforts. Considerations towards data protection and anonymisation varied across centres, from pseudonymised Personal Data to fully anonymised data, the latter being out of GDPR’s scope. Conclusion Unlike in clinical trials, multi-centre studies using real world data benefit from larger patient diversity. However, data collection can be challenging in late toxicities such as ORN if previous patient consent was not obtained. The different attitudes towards data sharing and acceptability of the type of DTA across all centres prolonged the negotiation between legal teams. We recommend initiating the legal process for the DTAs at an early stage, during protocol development and local data collection, to reduce delays. GDPR contains mechanisms and requirements for sharing Personal Data within and outside the EU, but consistency is lacking. Standardisation of the data sharing process to facilitate multi-centre collaborations would reduce unnecessary delays and costly resources and promote more robust clinical research studies, thus resulting in improved quality of patients care. S. Volpe 1 , F. Mastroleo 2 , A. Gaeta 3 , M. Zaffaroni 1 , M. Pepa 1 , M.G. Vincini 1 , S. Raimondi 3 , L.J. Isaksson 1 , C. Rampinelli 4 , M. Cremonesi 5 , S. Gandini 3 , M. Guckenberger 6 , R. Orecchia 7 , B.A. Jereczek-Fossa 1 1 Istituto Europeo di Oncologia IRCCS, Department of Radiation Oncology, Milan, Italy; 2 Università del Piemonte Orientale, Department of Translational Medicine, Novara, Italy; 3 Istituto Europeo di Oncologia IRCCS, Department of Experimental Oncology, Milan, Italy; 4 Istituto Europeo di Oncologia IRCCS, Department of Radiology, Milan, Italy; 5 Istituto Europeo di Oncologia IRCCS, Radiation Research Unit, Milan, Italy; 6 University Hospital Zurich, Department of Radiation Oncology, Zurich, Switzerland; 7 Istituto Europeo di Oncologia IRCCS, Scientific Directorate, Milan, Italy Purpose or Objective While 4D-computed tomography (CT) simulation represents a gold standard in stereotactic body radiotherapy (SBRT) for early-stage non-small cell lung cancer (ES-NSCLC), dedicated investigations on these images in radiomic studies are limited. This work aims to test the coefficient of variation (COV) of radiomic features across 10 respiratory phases and after applying different filtering methods. Materials and Methods Seventy 4D-CTs acquired with the same scanner and acquisition parameters at a single Institution were retrieved. Pre- processing and features extraction were implemented using Pyradiomics v3.0.1. Features were subdivided into 7 classes, namely first order, gray level co-occurrence matrix, gray level dependence matrix, gray level run length matrix, gray level zone matrix, neighboring gray tone difference matrix and shape. Null features in more than 90% of the cases in all respiratory phases were excluded. For each feature, the COV between the ten phases measurements was calculated for each patient [COV = (standard deviation/average) × 100]. Each feature is then represented by the average COVs among patients. The average COV was then classified as ≤ 5%, 5%20%. Results Almost 2000 features, mainly classified within the gray level co-occurrence matrix category, were extracted. Considering pre-processing methods, the majority of features derived from the wavelet (all permutations), lpb-3D and log-sigma filters (n= 744, 279 and 279, respectively). Qualitatively, COVs> 20% were observed across all categories and filters. Specifically, COVs>20% were more the most frequent in the gray level zone matrix and in the neighboring gray tone difference matrix categories. The application of pre-processing determined a different distribution of COVs, with a predominance of stable features (COVs ≤ 5%) in the lpb-2D and lpb-3D methods, while the largest variability was observed when the logarithm and log-sigma filters were used (Fig1). PO-2108 Impact of breathing and image filtering on radiomic features derived from 4D-CT in early-STAGE NSCLC