Abstract Book

S280

ESTRO 37

white matter (WM) for each patient from G1 images. Tumor volume (GTV) was manually segmented for G2 patients. Postprocessing was performed using an in-home Python code on G2 images (called G2’ after this step). Intensities were rescaled first to [0−32767] to erase inter- patient differences in intensity range. G2’ images were then resampled on a 0.5x0.5x0.5mm3 grid size. 8 histogram-based and 31 textural features were extracted using LIFEx freeware (Fig.1). Univariate analysis was performed in WM to assess the robustness of features between different datasets (Wilcoxon’s test: p<0.05 means feature value depends on MR scanner). Different methods of discretization were investigated for the calculation of textural features in GTV: relative (RD) or absolute (AD) with 32, 64, 128, 256 levels. Spearman correlation coefficients ρ were computed in GTV between all features, and between each feature and tumor volume.

Conclusion We present a renormalization of the Haralick features that are asymptotically invariant to the number of GLs in the image. The renormalized features enable statistical models to (I): be constructed from ROIs with varying quantization levels or (II): be applied to ROIs with different quantizaion levels. PV-0528 An innovative MRI harmonization method allowing large scale radiomics analysis in glioblastoma S. Reuzé 1,2,3 , A.S. Dirand 1,3 , R. Sun 2,3,4 , F. Orlhac 5 , G. Louvel 4 , S. Ammari 6 , E. Deutsch 2,3,4 , C. Robert 1,2,3 1 Gustave Roussy, Radiotherapy Department- Medical Physics Unit, Villejuif, France 2 INSERM U1030, Molecular Radiotherapy, Villejuif, France 3 Paris-Saclay University, Faculty of Medicine, Le Kremlin- Bicêtre, France 4 Gustave Roussy, Radiotherapy Department, Villejuif, France 5 CEA-SHFJ, IMIV, Orsay, France 6 Gustave Roussy, Medical Imaging Department, Villejuif, France Purpose or Objective Magnetic Resonance Imaging (MRI) is widely used during glioblastoma (GB) treatment. Quantitative features can be computed through radiomic analysis. In particular, textural features showed a great correlation with intra- tumor heterogeneity and treatment outcome, but also with acquisition parameters and type of MR scanner, so that most published studies are generally performed on small cohorts on a single device. The aim of our study was to develop a harmonization process to allow radiomic analysis of multicenter cohorts and strengthen statistical relevance of predictive models by external validation. Then, we assessed the robustness of radiomic features computed on multicenter images after post-processing step. Material and Methods 190 GB patients were retrospectively included: G1, N=62 for methodology elaboration (5 MR scanners); G2, N=128 for clinical validation (11 MR scanners). Contrast- enhanced T1-weighted series were collected. Two homogeneous volumes of interest were delineated in the

Results Basic intensity characteristics (min, max, mean) extracted from WM on G1 highlighted a large dependence on MR scanner (p<0.05). RD with 32 or 64 bins limits this dependence for textural features (G1). Only two 1st- order features were robust among all devices, whereas 7 histogram and 7 textural features were robust between 1.5T and 3T magnets (p>0.05, G1). Basic characteristics showed less variability on G2’ in GTV. RD highlighted correlations between features and tumor volume (|ρ|>0.5, 8 features). AD (256 bins in [0−32767]) decreases these correlations and reduces the dependence of features between devices (Fig.2). For 25 pairs of devices, there were at least 25 robust features, and 11 textural features were robust among magnetic field strength (p>0.05, G2’).