Computation of reliable textural indices from multimodal brain MRI: suggestions based on a study of patients with diffuse intrinsic pontine glioma

Jessica Goya-Outi,Agusti Alentorn,Nathalie Boddaert,Fanny Orlhac,Raphael Calmon,Stéphanie Puget,Jacques Grill,Irène Buvat,Vincent Frouin,Frederique Frouin,Cathy Philippe,Christophe Nioche

doi:10.1088/1361-6560/aabd21

Abstract

Few methodological studies regarding widely used textural indices robustness in MRI have been reported. In this context, this study aims to propose some rules to compute reliable textural indices from multimodal 3D brain MRI. Diagnosis and post-biopsy MR scans including T1, post-contrast T1, T2 and FLAIR images from thirty children with diffuse intrinsic pontine glioma (DIPG) were considered. The hybrid white stripe method was adapted to standardize MR intensities. Sixty textural indices were then computed for each modality in different regions of interest (ROI), including tumor and white matter (WM). Three types of intensity binning were compared : constant bin width and relative bounds; constant number of bins and relative bounds; constant number of bins and absolute bounds. The impact of the volume of the region was also tested within the WM. First, the mean Hellinger distance between patient-based intensity distributions decreased by a factor greater than 10 in WM and greater than 2.5 in gray matter after standardization. Regarding the binning strategy, the ranking of patients was highly correlated for 188/240 features when comparing with , but for only 20 when comparing with , and nine when comparing with . Furthermore, when using or texture indices reflected tumor heterogeneity as assessed visually by experts. Last, 41 features presented statistically significant differences between contralateral WM regions when ROI size slightly varies across patients, and none when using ROI of the same size. For regions with similar size, 224 features were significantly different between WM and tumor. Valuable information from texture indices can be biased by methodological choices. Recommendations are to standardize intensities in MR brain volumes, to use intensity binning with constant bin width, and to define regions with the same volumes to get reliable textural indices.

Highlights

The advent of radiomics generates a lot of enthusiasm and development, especially for oncological studies
MR intensity standardization The white matter (WM) and gray matter (GM) histograms before and after intensity standardization are shown in figure 5 for the four MR modalities
The mean Hellinger distance was decreased by a factor greater than 10 in WM, and by a factor greater than 2.5 in GM

Summary

Introduction

The advent of radiomics generates a lot of enthusiasm and development, especially for oncological studies. Radiomics helps in revealing tumor characteristics or predicting prognosis through the extraction of a great number of imaging indices inside the tumor area (Lambin et al 2012, Gevaert et al 2014, Aerts 2016, Gillies et al 2016, Skogen et al 2016) These indices are based on tumor shape, on first-order statistics, e.g. mean intensity, standard deviation, histogram-derived indices and on high-order statistics including textural indices derived. J Goya-Outi et al from the gray level co-occurrence matrix (GLCM) (Haralick et al 1973), the gray level run-length matrix (GLRLM) (Tang 1998) or the gray level size-zone matrix (GLSZM) (Thibault et al 2009) These textural indices have been widely used in medical imaging, their variability as a function of the range and binning of image intensity and the importance of the volume of the region of interest in which they are computed has not been extensively studied, especially in MRI. Among the possible sources of variation, it was previously shown that the volume of the region of interest (Orlhac et al 2014, Hatt et al 2015) and the intensity binning method (Leijenaar et al 2015, Orlhac et al 2015) play an important role

Objectives

Methods

Results

Discussion

Conclusion