Robustness of radiomic features in CT images with different slice thickness, comparing liver tumour and muscle

Eva Mendes Serrao,Leonardo Rundo,Andrew B. Gill,Evis Sala,Matthew Hoare,Lorena Escudero Sanchez

doi:10.1038/s41598-021-87598-w

Abstract

Radiomic image features are becoming a promising non-invasive method to obtain quantitative measurements for tumour classification and therapy response assessment in oncological research. However, despite its increasingly established application, there is a need for standardisation criteria and further validation of feature robustness with respect to imaging acquisition parameters. In this paper, the robustness of radiomic features extracted from computed tomography (CT) images is evaluated for liver tumour and muscle, comparing the values of the features in images reconstructed with two different slice thicknesses of 2.0 mm and 5.0 mm. Novel approaches are presented to address the intrinsic dependencies of texture radiomic features, choosing the optimal number of grey levels and correcting for the dependency on volume. With the optimal values and corrections, feature values are compared across thicknesses to identify reproducible features. Normalisation using muscle regions is also described as an alternative approach. With either method, a large fraction of features (75–90%) was found to be highly robust (< 25% difference). The analyses were performed on a homogeneous CT dataset of 43 patients with hepatocellular carcinoma, and consistent results were obtained for both tumour and muscle tissue. Finally, recommended guidelines are included for radiomic studies using variable slice thickness.

Highlights

Radiomic image features are becoming a promising non-invasive method to obtain quantitative measurements for tumour classification and therapy response assessment in oncological research
Unlike qualitative image evaluation, which requires a trained reader to make a subjective judgement based on images, radiomics allows a large number of quantitative features to be extracted from standard-of-care images, from modalities such as computed tomography (CT), Magnetic Resonance Imaging (MRI) and Positron Emission Tomography (PET)
Most of the radiomic features were found to have a correlation or dependency on either the number of grey levels (GLs) or the number of voxels. Some features presented both dependencies. Results evaluating those intrinsic dependencies are discussed for the texture radiomic features: first-order, Grey Level Co-occurrence Matrix features (GLCM), Grey Level Dependence Matrix (GLDM), Grey Level Run Length Matrix (GLRLM), Grey Level Size Zone Matrix (GLSZM), Neighbouring Grey Tone Difference Matrix Features (NGTDM)

Summary

Introduction

Radiomic image features are becoming a promising non-invasive method to obtain quantitative measurements for tumour classification and therapy response assessment in oncological research. Unlike qualitative image evaluation, which requires a trained reader to make a subjective judgement based on images (i.e. presence of disease), radiomics allows a large number of quantitative features to be extracted from standard-of-care images, from modalities such as CT, MRI and Positron Emission Tomography (PET). These features provide quantitative measurements of tissue characteristics, such as shape or heterogeneity, allowing objective, reader-independent non-invasive biomarkers to be extracted. To make practical choices—such as decisions on voxel size and quantisation of grey levels (GLs), necessary to obtain robust and reliable o utcomes

Objectives

Methods

Results

Conclusion