Robustness of Radiomic Features in 68Ga-DOTATOC PET/CT: A Monocentric Experience for Neuroendocrine Tumors

#2973

Introduction: Neuroendocrine tumors (NET) are a heterogeneous group of malignancies with varied histology, localization and behaviour.

Aim(s): In order to use 68Ga-DOTATOC PET/CT as a prognostic and predictive tool for tumoral heterogeneity in NET, it is important to identify radiomic features (RFs) that must be repeatable and reliable as a function of segmentation methods and post-processing sets.

Materials and methods: We retrospectively analyzed data from 71 patients, who perform a PET/CT scan with 68Ga-DOTATOC for NET, to evaluate the robustness of RFs in function of segmentation methods. Contouring was performed on a total of 116 lesions with two different contouring methods: manual contours using the software LIFEx by two operators and semi-automatic edge-based method using MATLAB homemade code. For each region of interest (ROIs), SUVmax fixed thresholds (20, 30 and 40%) were applied and three different intensity rescale factors were employed (0-60, 0-80, min-max of the SUV of the ROI). 45 RFs were extracted using LIFEx. The impact of the different parameters analyzed on RFs was assessed by Intra-class Correlation Coefficients (ICC) using statistical R software. A RF was considered robust if ICC was >0.8.

Conference: 17th Annual ENETSConcerence (2020)

Presenting Author: Liberini V

Authors: Liberini V, De Santi B, Gallio E, Rampado O, Dionisi B,

Keywords: NET, heterogeneous, 68Ga-DOTATOC, contouring methods, LIFEx, radiomic, texture features, PET/CT,

To read the full abstract, please log into your ENETS Member account.