Impact of image preprocessing on the scanner dependence of multi-parametric MRI radiomic features and covariate shift in multi-institutional glioblastoma datasets

Abstract

Recent advances in radiomics have enhanced the value of medical imaging in various aspects of clinical practice, but a crucial component that remains to be investigated further is the robustness of quantitative features to imaging variations and across multiple institutions. In the case of MRI, signal intensity values vary according to the acquisition parameters used, yet no consensus exists on which preprocessing techniques are favorable in reducing scanner-dependent variability of image-based features. Hence, the purpose of this study was to assess the impact of common image preprocessing methods on the scanner dependence of MRI radiomic features in multi-institutional glioblastoma multiforme (GBM) datasets. Two independent GBM cohorts were analyzed: 50 cases from the TCGA-GBM dataset and 111 cases acquired in our institution, and each case consisted of 3 MRI sequences viz. FLAIR, T1-weighted, and T1-weighted post-contrast. Five image preprocessing techniques were examined: 8-bit global rescaling, 8-bit local rescaling, bias field correction, histogram standardization, and isotropic resampling. A total of 420 features divided into eight categories representing texture, shape, edge, and intensity histogram were extracted. Two distinct imaging parameters were considered: scanner manufacturer and scanner magnetic field strength. Wilcoxon tests identified features robust to the considered acquisition parameters under the selected image preprocessing techniques. A machine learning-based strategy was implemented to measure the covariate shift between the analyzed datasets using features computed using the aforementioned preprocessing methods. Finally, radiomic scores (rad-scores) were constructed by identifying features relevant to patients' overall survival after eliminating those impacted by scanner variability. These were then evaluated for their prognostic significance through Kaplan-Meier and Cox hazards regression analyses. Our results demonstrate that overall, histogram standardization contributes the most in reducing radiomic feature variability as it is the technique to reduce the covariate shift for three feature categories and successfully discriminate patients into groups of different survival risks.

Impact of image preprocessing methods on reproducibility of radiomic features in multimodal magnetic resonance imaging in glioblastoma

“…Variation in the brain MRI intensity distribution because of the difference in inter‐ and intra‐scanners sensitivity and acquisition parameters leads to complicated in image quantitative analysis, even for the same protocol, tissue, patient, and scanner. Consequently, intensity normalization is essential for providing the same tissue intensity scale in brain MR images across all observations to facilitate radiomics analysis and accurate quantitative comparison between MRI volumes as is expressed by Eq. (), are reference image and the original image intensity respectively, and are respectively the low and high reference image intensity range.…”

“…Given that MRI undergoes of various inherent acquisition artifacts and noises such as lack of standard intensity for inter‐ and intra‐scanner variability even for the same protocol, body region, and patient; intensity non‐uniformity as a result of reduced radio frequency, coil uniformity, nonlinear fields, gradient field, magnetic field, and etc. ; image preprocessing method suchlike intensity normalization, bias field correction, and noise smoothing can facilitate quantitative MRI analysis and make the radiomics results more repeatable and comparable . Currently, many attentions of the GBM mMRI‐based radiomics studies were drowned to prognosis and prediction model, while they have not used a pre‐specific image preprocessing pipeline.…”

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

“…Even if several studies have shown variabilities in texture analysis depending on MRI acquisition parameters and the grey-level discretization step, none of them has assessed the combined impact of intensity normalization and grey-level discretization pre-processing methods on radiomics feature values in MRI [36][37][38][39][40] .…”

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

“…The need for some preprocessing steps before extracting MRI radiomic features has been very recently acknowledged for other tumor types (41,42). Although there is no consensus on the preprocessing methods that should be used, two main pitfalls that are specific to MRI have been identified.…”