Prediction of cervical cancer recurrence using textural features extracted from 18F-FDG PET images acquired with different scanners

// Sylvain Reuzé 1, 2, 3, 4 , Fanny Orlhac 1, 5 , Cyrus Chargari 1, 2, 3, 6, 7 , Christophe Nioche 5 , Elaine Limkin 1 , François Riet 3 , Alexandre Escande 3 , Christine Haie-Meder 3 , Laurent Dercle 8, 9 , Sébastien Gouy 10 , Irène Buvat 5 , Eric Deutsch 1, 2, 3 and Charlotte Robert 1, 2, 3, 4 1 INSERM, U1030, F-94805, Villejuif, France 2 Université Paris-Sud, Université Paris-Saclay, F-94270, Le Kremlin-Bicêtre, France 3 Gustave Roussy, Université Paris-Saclay, Department of Radiotherapy, F-94805, Villejuif, France 4 Gustave Roussy, Université Paris-Saclay, Department of Medical Physics, F-94805, Villejuif, France 5 IMIV, CEA, INSERM, CNRS, Université Paris-Sud, Université Paris-Saclay, CEA-SHFJ, Orsay, France 6 French Military Health Services Academy, Ecole du Val-de-Grâce, Paris, France 7 Institut de Recherche Biomédicale des Armées, Bretigny-sur-Orge, France 8 INSERM, U1015, F-94805, Villejuif, France 9 Gustave Roussy, Université Paris-Saclay, Department of Nuclear Medicine and Endocrine Oncology, F-94805, Villejuif, France 10 Gustave Roussy, Université Paris-Saclay, Department of Gynecologic Surgery, F-94805, Villejuif, France Correspondence to: Charlotte Robert, email: ch.robert@gustaveroussy.fr Keywords: radiomics, cervical cancer, texture, PET imaging Received: February 24, 2017     Accepted: April 11, 2017     Published: May 15, 2017 ABSTRACT Objectives: To identify an imaging signature predicting local recurrence for locally advanced cervical cancer (LACC) treated by chemoradiation and brachytherapy from baseline 18 F-FDG PET images, and to evaluate the possibility of gathering images from two different PET scanners in a radiomic study. Methods: 118 patients were included retrospectively. Two groups (G1, G2) were defined according to the PET scanner used for image acquisition. Eleven radiomic features were extracted from delineated cervical tumors to evaluate: (i) the predictive value of features for local recurrence of LACC, (ii) their reproducibility as a function of the scanner within a hepatic reference volume, (iii) the impact of voxel size on feature values. Results: Eight features were statistically significant predictors of local recurrence in G1 (p < 0.05). The multivariate signature trained in G2 was validated in G1 (AUC=0.76, p<0.001) and identified local recurrence more accurately than SUV max (p=0.022). Four features were significantly different between G1 and G2 in the liver. Spatial resampling was not sufficient to explain the stratification effect. Conclusion: This study showed that radiomic features could predict local recurrence of LACC better than SUV max . Further investigation is needed before applying a model designed using data from one PET scanner to another.