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

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 [Formula: see text]: constant bin width and relative bounds; [Formula: see text] constant number of bins and relative bounds; [Formula: see text] 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 [Formula: see text] with [Formula: see text], but for only 20 when comparing [Formula: see text] with [Formula: see text], and nine when comparing [Formula: see text] with [Formula: see text]. Furthermore, when using [Formula: see text] or [Formula: see text] 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.

Standardization of brain MR images across machines and protocols: bridging the gap for MRI-based radiomics

“…Goya-Outi et al . investigated the impact of intensity binning combined with WhiteStripe normalization on 30 patients suffering from diffuse intrinsic pontine glioma 44 . They compared patient ranking based on radiomics features to visual assessment of the heterogeneity.…”

Gray-level discretization impacts reproducible MRI radiomics texture features

“…They found that modifying the discretization parameters led to significantly different results. More recently, Goya-Outi et al [14] investigated the impact of the absolute and relative discretization methods on patient ranking compared to the visual assessment of the heterogeneity of diffuse intrinsic pontine gliomas on 30 patients and 4 MR sequences (T1-WI, post-contrast T1-WI, T2-WI and Fluid Attenuation Inversion Recovery (FLAIR)). They concluded that using an absolute (FBS) discretization method provided more consistent results compared to the visual assessment of the heterogeneity.…”

Correction for Magnetic Field Inhomogeneities and Normalization of Voxel Values Are Needed to Better Reveal the Potential of MR Radiomic Features in Lung Cancer

“…Significant variations in mean intensity values measured in similar tissues (for instance subcutaneous fat for lung studies, white matter for brain studies) can be observed between different patients even when using a similar acquisition protocol on the same scanner (26). These variations are a major pitfall for radiomic studies (26). The intensity normalization aims at reducing the intensity variations between different patients.…”

“…Still, the use of different scanners and/or different image acquisition and reconstruction protocols introduces some variability in voxel values hence in radiomic features (24) and some harmonization techniques have been proposed to realign radiomic features measured in different conditions (25). In anatomical MR sequences, images are initially expressed in arbitrary units, meaning that a given tissue type (for instance fat) will not always yield a similar voxel value, even when the images are acquired in the same patient and same conditions using the same scanner (26). Therefore, a measured voxel value cannot be readily interpreted in terms of well-understood physics quantity unlike in CT (a HU does correspond to a unique attenuation coefficient) and in PET (an SUV corresponds to a unique tracer concentration in a given patient).…”

“…MRI preprocessing, posterior to acquisitions, including magnetic bias field correction and normalization has been successfully applied to brain studies using white matter as the reference tissue (26,27). This approach has been rarely applied to other organs.…”