ESTRO 38 Abstract book

S1030 ESTRO 38

EP-1895 Multicenter CT phantoms public dataset for radiomics reproducibility studies P. Kalendralis 1 , A. Traverso 1 , Z. Shi 1 , I. Zhovannik 1,2 , R. Monshouwer 2 , M.P.A. Starmans 3 , S. Klein 3,4 , P. Elisabeth 5 , R. Boellaard 6 , A. Dekker 1 , L. Wee 1 1 Maastro Clinic, Department of Radiation Oncology Maastro- GROW School for Oncology and Developmental Biology- Maastricht University Medical Centre+- Maastricht- The Netherlands, Maastricht, The Netherlands ; 2 Radboud University Medical Center- Nijmegen- The Netherlands, Department of Radiation Oncology, Nijmegen, The Netherlands ; 3 Erasmus Medical Centre- Rotterdam- The Netherlands, Department of Radiology and Nuclear Medicine, Rotterdam, The Netherlands ; 4 Erasmus Medical Centre- Rotterdam- The Netherlands, Department of Medical Informatics, Rotterdam, The Netherlands ; 5 University Medical Center Groningen- Groningen- The Netherlands, University Medical Center Groningen- Groningen- The Netherlands, Groningen, The Netherlands ; 6 VU Medical Center- Amsterdam- The Netherlands., Department of Radiology and Nuclear Medicine- VU Medical Center- Amsterdam- The Netherlands., Amsterdam, The Netherlands Purpose or Objective Reproducibility is a fundamental requirement for radiomics-based clinical prediction models. The aim of this study is to describe and provide an open-access Computed Tomography (CT) phantoms image dataset acquired at three different Dutch centers. The data was acquired for radiomics reproducibility studies with respect to variations in acquisition settings, scanners and reconstruction algorithms. Material and Methods The CT images of the Catphan 700, COPDGene Phantom II (Phantom Laboratory, Greenwich, NY, USA), and the Triple-modality 3D Abdominal Phantom (CIRS, Norfolk, Virginia, USA) were exported in Digital Imaging and Communications in Medicine (DICOM) format from Siemens and Philips scanners. Participating centres were MAASTRO Clinic (Maastricht, NL), Radboud University Medical Center (Nijmegen, NL), and University Medical Center Groningen (Groningen, NL). The scans were acquired by varying the parameters of slice thickness, reconstruction kernels, and tube current. The regions of interest (ROIs) of the COPDGene and Abdominal phantoms were delineated manually in MIRADA DBx (version 1.2.0.59, Mirada Medical, Oxford, United Kingdom). The Catphan700 was used for quality assessment of the scans. Representative axial slices of phantoms and the delineated ROIs of the scans are shown in Figure 1. Results The scans are publically accessible and reusable via an instance of the eXtensible Neuroimaging Archive Toolkit (XNAT) hosted within the national Dutch research infrastructure (TraIT, www.ctmm-trait.nl). The Phantom Laboratory images have been uploaded to the XNAT collection STW-STRATEGY-Phantom_Series1: (https://xnat.bmia.nl/data/projects/stwstrategyps1). The CIRS multimodality Abdominal Phantom images have been uploaded to the SNAT collection STW-STRATEGY- Phantom_Series2: (https://xnat.bmia.nl/data/projects/stwstrategyps2). An overview of the parameters and protocols used is presented in Table 1. Conclusion The phantoms dataset are offered to the radiomics community for the comparison of simple features extracted with different software pipelines. The dataset could be potentially useful for the identification of unstable features with respect to different scanning and acquisition parameters.

Figure 1. The three different phantoms used in the multicenter study with the delineated spherical ROIs.

Table 1. Imaging acquisition parameters and protocols used for the baseline scans across the three different scanners. EP-1896 A TCP-based early-regression index predicts outcome of rectal cancer patients better than pCR C. Fiorino 1 , P. Passoni 2 , C. Gumina 2 , A. Palmisano 3 , S. Broggi 1 , G.M. Cattaneo 1 , A. Di Chiara 3 , M. Mori 1 , R. Rosati 4 , N. Slim 2 , F. De Cobelli 3 , R. Calandrino 1 , N.G. Di Muzio 2 1 San Raffaele Scientific Institute, Medical Physics, Milano, Italy ; 2 San Raffaele Scientific Institute, Radiotherapy, Milano, Italy ; 3 San Raffaele Scientific Institute, Radiology, Milano, Italy ; 4 San Raffaele Scientific Institute, Gastroenterology Surgery, Milano, Italy Purpose or Objective A radiobiological parameter (early regression index, ERI TCP ) based on early tumor (GTV) regression was previously found to predict pathological complete remission (pCR) after neo-adjuvant radio-chemotherapy (RCT) of locally advanced rectal cancer (Rca). In current investigation, ERI TCP was tested against conventional clinical and pathological biomarkers in predicting long- term outcome after RCT delivered within an observational study testing an early regression- guided adaptive boosting technique (ART).