Abstract

Info

Title:: An analysis of influence of different equipment and different scanning parameters on reproducibility of radiomics features

Author(s):: ZHANG Bailin; DAI Zhenhui; ZHU Lin; YANG Geng; ZHU Yuanhu; ZHAO Shiwu; LI Fei; CAI Chunya; WANG Xuetao; Radiotherapy Center, the Second Affiliated Hospital of Guangzhou University of Chinese Medicine, Guangzhou, 510006, China

Keywords:: Keywords: radiomics reproducibility computed tomography scanning parameter

Abstract:: Abstract: Objective To analyse the reproducibility of radiomics features in different CT equipment with various scanning parameters. Methods Different scanning parameters were set on three CT machines: different reconstruction slice thickness (1, 3, 5 mm respectively), different reconstruction kernel function (related to specific CT machines), different exposure time (50, 100, 300 mAs respectively), and two kinds of phantom were scanned to extract radiomics features and evaluate their reproducibility. Results 81.6% of the features had poor reproducibility (CV>10%) under different slice thickness, and only 8.8% of the features in the three machines met the requirements (CV<10%) 46.4% of the features have poor reproducibility under different reconstruction kernel functions, and only 12.1% of the features in three machines meet the requirements (CV<10%) 31.7% of the features met the reproducibility requirements only in a single machine 54% of the features have poor reproducibility under different exposure time. Conclusion When using radiomics to establish a model to study clinical problems, the adverse effects of scanning parameters on the reproducibility of imaging histology features should be fully considered and avoided in order to obtain better clinical research results.