Toward a Method of Selecting Robust Heterogeneous PET Images Radiomic Features

More recent attention has focused on the provision of texture analysis for quantification of intratumor uptake heterogeneity in PET/CT images. This allows the extraction of quantitative features from medical images now termed ‘radiomic’ to become a biomarker and promising results have correlated these features with end point information (i.e. tumor type, therapy response, prognosis). Most of previous studies provide important insights into the robustness of radiomic features against different type of protocols and conditions. However, the ability of these robust features to identify the differences between regions (inserts) still unknown. Using advanced analysis techniques for feature selection can be serve as a promising method to alleviate redundancy in radiomics. The implementation of the Friedman test to radiomic analysis serves as a powerful exploratory instrument to reveal the ability of texture of distinguishing between heterogeneous tumors. For this purpose, texture features from PET-CT images of phantom with 4 tumors inserts with different level of heterogeneity were analyzed using the Friedman test to assess the ability of radiomic features to capture heterogeneity differences.