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

Radiomics relies on the extraction of a wide variety of quantitative image-based features to provide decision support. Magnetic resonance imaging (MRI) contributes to the personalization of patient care but suffers from being highly dependent on acquisition and reconstruction parameters. Today, there are no guidelines regarding the optimal pre-processing of MR images in the context of radiomics, which is crucial for the generalization of published image-based signatures. This study aims to assess the impact of three different intensity normalization methods (Nyul, WhiteStripe, Z-Score) typically used in MRI together with two methods for intensity discretization (fixed bin size and fixed bin number). The impact of these methods was evaluated on first- and second-order radiomics features extracted from brain MRI, establishing a unified methodology for future radiomics studies. Two independent MRI datasets were used. The first one (DATASET1) included 20 institutional patients with WHO grade II and III gliomas who underwent post-contrast 3D axial T1-weighted (T1w-gd) and axial T2-weighted fluid attenuation inversion recovery (T2w-flair) sequences on two different MR devices (1.5 T and 3.0 T) with a 1-month delay. Jensen–Shannon divergence was used to compare pairs of intensity histograms before and after normalization. The stability of first-order and second-order features across the two acquisitions was analysed using the concordance correlation coefficient and the intra-class correlation coefficient. The second dataset (DATASET2) was extracted from the public TCIA database and included 108 patients with WHO grade II and III gliomas and 135 patients with WHO grade IV glioblastomas. The impact of normalization and discretization methods was evaluated based on a tumour grade classification task (balanced accuracy measurement) using five well-established machine learning algorithms. Intensity normalization highly improved the robustness of first-order features and the performances of subsequent classification models. For the T1w-gd sequence, the mean balanced accuracy for tumour grade classification was increased from 0.67 (95% CI 0.61–0.73) to 0.82 (95% CI 0.79–0.84, P = .006), 0.79 (95% CI 0.76–0.82, P = .021) and 0.82 (95% CI 0.80–0.85, P = .005), respectively, using the Nyul, WhiteStripe and Z-Score normalization methods compared to no normalization. The relative discretization makes unnecessary the use of intensity normalization for the second-order radiomics features. Even if the bin number for the discretization had a small impact on classification performances, a good compromise was obtained using the 32 bins considering both T1w-gd and T2w-flair sequences. No significant improvements in classification performances were observed using feature selection. A standardized pre-processing pipeline is proposed for the use of radiomics in MRI of brain tumours. For models based on first- and second-order features, we recommend normalizing images with the Z-Score method and adopting an absolute discretization approach. For second-order feature-based signatures, relative discretization can be used without prior normalization. In both cases, 32 bins for discretization are recommended. This study may pave the way for the multicentric development and validation of MR-based radiomics biomarkers.

[1]  Baojuan Li,et al.  Radiomics Strategy for Molecular Subtype Stratification of Lower‐Grade Glioma: Detecting IDH and TP53 Mutations Based on Multimodal MRI , 2018, Journal of magnetic resonance imaging : JMRI.

[2]  M. Hatt,et al.  Intratumor Heterogeneity Characterized by Textural Features on Baseline 18F-FDG PET Images Predicts Response to Concomitant Radiochemotherapy in Esophageal Cancer , 2011, The Journal of Nuclear Medicine.

[3]  Suman Sarkar,et al.  Comprehensive analysis of single molecule sequencing-derived complete genome and whole transcriptome of Hyposidra talaca nuclear polyhedrosis virus , 2018, Scientific Reports.

[4]  Paul Kinahan,et al.  Radiomics: Images Are More than Pictures, They Are Data , 2015, Radiology.

[5]  Christos Davatzikos,et al.  Multi-stage Association Analysis of Glioblastoma Gene Expressions with Texture and Spatial Patterns , 2018, BrainLes@MICCAI.

[6]  Anant Madabhushi,et al.  Interplay between bias field correction, intensity standardization, and noise filtering for T2-weighted MRI , 2011, 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[7]  Ginu A. Thomas,et al.  A Coclinical Radiogenomic Validation Study: Conserved Magnetic Resonance Radiomic Appearance of Periostin-Expressing Glioblastoma in Patients and Xenograft Models , 2018, Clinical Cancer Research.

[8]  Jakob Nikolas Kather,et al.  CD163+ immune cell infiltrates and presence of CD54+ microvessels are prognostic markers for patients with embryonal rhabdomyosarcoma , 2019, Scientific Reports.

[9]  Geoffrey G. Zhang,et al.  Intrinsic dependencies of CT radiomic features on voxel size and number of gray levels , 2017, Medical physics.

[10]  C. Crainiceanu,et al.  Statistical normalization techniques for magnetic resonance imaging , 2014, NeuroImage: Clinical.

[11]  P. Huang,et al.  Distinguishing True Progression From Radionecrosis After Stereotactic Radiation Therapy for Brain Metastases With Machine Learning and Radiomics. , 2018, International journal of radiation oncology, biology, physics.

[12]  Tobias Gauer,et al.  Radiomics of Brain MRI: Utility in Prediction of Metastatic Tumor Type. , 2019, Radiology.

[13]  Vincent Frouin,et al.  Computation of reliable textural indices from multimodal brain MRI: suggestions based on a study of patients with diffuse intrinsic pontine glioma , 2018, Physics in medicine and biology.

[14]  N. Dogan,et al.  Quantitative Radiomics: Impact of Pulse Sequence Parameter Selection on MRI-Based Textural Features of the Brain , 2018, Contrast media & molecular imaging.

[15]  El Naqa,et al.  A radiomics model from joint FDG-PET and MRI texture features for the prediction of lung metastases in soft-tissue sarcomas of the extremities , 2015 .

[16]  A. Giannakopoulos,et al.  Dynamic behavior of suture-anastomosed arteries and implications to vascular surgery operations , 2015, Biomedical engineering online.

[17]  Aaron Carass,et al.  Evaluating the Impact of Intensity Normalization on MR Image Synthesis , 2018, Image Processing.

[18]  Seok-Gu Kang,et al.  Radiomic MRI Phenotyping of Glioblastoma: Improving Survival Prediction. , 2018, Radiology.

[19]  Lin Shi,et al.  Histogram-based normalization technique on human brain magnetic resonance images from different acquisitions , 2015, Biomedical engineering online.

[20]  Luca Mainardi,et al.  Technical Note: Virtual phantom analyses for preprocessing evaluation and detection of a robust feature set for MRI‐radiomics of the brain , 2019, Medical physics.

[21]  A Lecler,et al.  Combining Multiple Magnetic Resonance Imaging Sequences Provides Independent Reproducible Radiomics Features , 2019, Scientific Reports.

[22]  Harini Veeraraghavan,et al.  Reliability of tumor segmentation in glioblastoma: impact on the robustness of MRI-radiomic features. , 2019, Medical physics.

[23]  Christos Davatzikos,et al.  Advancing The Cancer Genome Atlas glioma MRI collections with expert segmentation labels and radiomic features , 2017, Scientific Data.

[24]  Qiyong Gong,et al.  Automated Machine Learning Based on Radiomics Features Predicts H3 K27M Mutation in Midline Gliomas of the Brain. , 2019, Neuro-oncology.

[25]  Andriy Fedorov,et al.  Computational Radiomics System to Decode the Radiographic Phenotype. , 2017, Cancer research.

[26]  Tamim Niazi,et al.  Radiomics in Glioblastoma: Current Status and Challenges Facing Clinical Implementation , 2019, Front. Oncol..

[27]  Zev A. Binder,et al.  Radiomic MRI signature reveals three distinct subtypes of glioblastoma with different clinical and molecular characteristics, offering prognostic value beyond IDH1 , 2018, Scientific Reports.

[28]  Lian-pan Wu,et al.  Treatment response prediction of rehabilitation program in children with cerebral palsy using radiomics strategy: protocol for a multicenter prospective cohort study in west China. , 2019, Quantitative imaging in medicine and surgery.

[29]  Stephen M Smith,et al.  Fast robust automated brain extraction , 2002, Human brain mapping.

[30]  Adolf Pfefferbaum,et al.  The SRI24 multichannel atlas of normal adult human brain structure , 2009, Human brain mapping.

[31]  D. Ben Bashat,et al.  Differentiation between glioblastoma, brain metastasis and subtypes using radiomics analysis , 2019, Journal of magnetic resonance imaging : JMRI.

[32]  P. Lambin,et al.  Radiomics: the bridge between medical imaging and personalized medicine , 2017, Nature Reviews Clinical Oncology.

[33]  A Vamvakas,et al.  Imaging biomarker analysis of advanced multiparametric MRI for glioma grading. , 2019, Physica medica : PM : an international journal devoted to the applications of physics to medicine and biology : official journal of the Italian Association of Biomedical Physics.

[34]  T. Jiang,et al.  A radiomic signature as a non-invasive predictor of progression-free survival in patients with lower-grade gliomas , 2018, NeuroImage: Clinical.

[35]  Wei Chen,et al.  Establishment and ultrasound characteristics of atherosclerosis in rhesus monkey , 2015, Biomedical engineering online.

[36]  Stephen M. Moore,et al.  The Cancer Imaging Archive (TCIA): Maintaining and Operating a Public Information Repository , 2013, Journal of Digital Imaging.

[37]  Shuang Wu,et al.  Radiomics-based machine learning methods for isocitrate dehydrogenase genotype prediction of diffuse gliomas , 2019, Journal of Cancer Research and Clinical Oncology.

[38]  Robert J. Gillies,et al.  PO-0968: The effect of SUV discretization in PET Radiomics: the need for standardized methodology in tumor texture analysis , 2015 .

[39]  Jianhua Lin,et al.  Divergence measures based on the Shannon entropy , 1991, IEEE Trans. Inf. Theory.

[40]  Christopher. Simons,et al.  Machine learning with Python , 2017 .

[41]  David Moratal,et al.  Identification of the presence of ischaemic stroke lesions by means of texture analysis on brain magnetic resonance images , 2019, Comput. Medical Imaging Graph..

[42]  B. Ang,et al.  Overall survival prediction in glioblastoma multiforme patients from volumetric, shape and texture features using machine learning. , 2018, Surgical oncology.

[43]  Harini Veeraraghavan,et al.  Impact of image preprocessing on the scanner dependence of multi-parametric MRI radiomic features and covariate shift in multi-institutional glioblastoma datasets , 2019, Physics in medicine and biology.

[44]  Brian B. Avants,et al.  N4ITK: Improved N3 Bias Correction , 2010, IEEE Transactions on Medical Imaging.

[45]  Daniel Balvay,et al.  Gray-level discretization impacts reproducible MRI radiomics texture features , 2019, PloS one.

[46]  Andrzej Materka,et al.  Effects of MRI acquisition parameter variations and protocol heterogeneity on the results of texture analysis and pattern discrimination: an application-oriented study. , 2009, Medical physics.

[47]  N. Paragios,et al.  Promises and challenges for the implementation of computational medical imaging (radiomics) in oncology , 2017, Annals of oncology : official journal of the European Society for Medical Oncology.

[48]  Estanislao Arana,et al.  Lack of robustness of textural measures obtained from 3D brain tumor MRIs impose a need for standardization , 2017, PloS one.

[49]  Gaël Varoquaux,et al.  Scikit-learn: Machine Learning in Python , 2011, J. Mach. Learn. Res..

[50]  S. Arridge,et al.  Sources of intensity nonuniformity in spin echo images at 1.5 T , 1994, Magnetic resonance in medicine.

[51]  L G Nyúl,et al.  On standardizing the MR image intensity scale , 1999, Magnetic resonance in medicine.

[52]  Anant Madabhushi,et al.  Empirical evaluation of cross-site reproducibility in radiomic features for characterizing prostate MRI , 2018, Medical Imaging.

[53]  M. Soussan,et al.  A Postreconstruction Harmonization Method for Multicenter Radiomic Studies in PET , 2018, The Journal of Nuclear Medicine.

[54]  Cheng Li,et al.  Adjusting batch effects in microarray expression data using empirical Bayes methods. , 2007, Biostatistics.

[55]  Hajar Moradmand,et al.  Impact of image preprocessing methods on reproducibility of radiomic features in multimodal magnetic resonance imaging in glioblastoma , 2019, Journal of applied clinical medical physics.

[56]  D. Louis Collins,et al.  Evaluating intensity normalization on MRIs of human brain with multiple sclerosis , 2011, Medical Image Anal..

[57]  Steffen Löck,et al.  Image biomarker standardisation initiative , 2016 .

[58]  Fanny Orlhac,et al.  Validation of A Method to Compensate Multicenter Effects Affecting CT Radiomics. , 2019, Radiology.

[59]  Li Zhao,et al.  Oroxylin a Inhibits the Protection of Bone Marrow Microenvironment on CML Cells Through CXCL12/CXCR4/P-gp Signaling Pathway , 2019, Front. Oncol..

[60]  Milan Hájek,et al.  Phantoms for texture analysis of MR images. Long-term and multi-center study. , 2004, Medical physics.

[61]  Ron Kikinis,et al.  Repeatability of Multiparametric Prostate MRI Radiomics Features , 2018, Scientific Reports.

[62]  N. Paragios,et al.  A radiomics approach to assess tumour-infiltrating CD8 cells and response to anti-PD-1 or anti-PD-L1 immunotherapy: an imaging biomarker, retrospective multicohort study. , 2018, The Lancet. Oncology.

[63]  F. Tixier,et al.  Preoperative MRI-radiomics features improve prediction of survival in glioblastoma patients over MGMT methylation status alone , 2019, Oncotarget.

[64]  Thomas Philip Runarsson,et al.  Test-retest reliability and feature selection in physiological time series classification , 2012, Comput. Methods Programs Biomed..

[65]  Johan Pallud,et al.  Imaging of gliomas at 1.5 and 3 Tesla - A comparative study. , 2015, Neuro-oncology.

[66]  H. Aerts,et al.  Applications and limitations of radiomics , 2016, Physics in medicine and biology.

[67]  R. Steenbakkers,et al.  The Image Biomarker Standardization Initiative: Standardized Quantitative Radiomics for High-Throughput Image-based Phenotyping. , 2020, Radiology.

[68]  Joachim M. Buhmann,et al.  The Balanced Accuracy and Its Posterior Distribution , 2010, 2010 20th International Conference on Pattern Recognition.

[69]  Boqiang Liu,et al.  Computer-Aided Grading of Gliomas Combining Automatic Segmentation and Radiomics , 2018, Int. J. Biomed. Imaging.

[70]  Anant Madabhushi,et al.  Radiomics Analysis on FLT-PET/MRI for Characterization of Early Treatment Response in Renal Cell Carcinoma: A Proof-of-Concept Study1 , 2016, Translational oncology.