Harmonizing multicenter quantitative imaging data: sources of variability, statistical solutions, and practical workflows in CT and MRI

Abstract

Multicenter imaging studies are increasingly critical in epidemiology, yet variability across scanners, acquisition protocols, and reconstruction algorithms introduces systematic biases that threaten reproducibility and comparability of quantitative biomarkers. This paper reviews the major sources of heterogeneity in MRI, CT, and PET-CT data, highlighting their impact on epidemiologic inference, including misclassification, reduced statistical power, and compromised generalizability. We outline harmonization strategies spanning pre-acquisition standardization, phantom-based calibration, post-acquisition intensity normalization, and advanced statistical and machine learning methods such as ComBat and domain adaptation. Illustrative examples from MRI flow quantification and radiomic feature extraction demonstrate how harmonization can mitigate site effects and enable robust large-scale analyses.

Keywords

biomarkers / imaging / multicenter studies / image processing / reproducibility

References

[1] Jovicich J, Czanner S, Han X, Salat D, van der Kouwe A, Quinn B, Pacheco J, et al. MRI-derived measurements of human subcortical, ventricular and intracranial brain volumes: Reliability effects of scan sessions, acquisition sequences, data analyses, scanner upgrade, scanner vendors and field strengths. Neuroimage. 2009; 46:177-92.

[2] Fortin J, Parker D, Tunç B, Watanabe T, Elliott MA, Ruparel K, et al. Harmonization of multi-site diffusion tensor imaging data. Neuroimage. 2017; 161:149-70.

[3] Das S, Zijdenbos AP, Harlap J, Vins D, Evans AC. LORIS: a web-based data management system for multi-center studies. Front Neuroinform. 2012; 5:37.

[4] Han X, Jovicich J, Salat D, van der Kouwe A, Quinn B, Czanner S, et al. Reliability of MRI-derived measurements of human cerebral cortical thickness: the effects of field strength, scanner upgrade and manufacturer. Neuroimage. 2006; 32:180-94.

[5] Reig S, Sánchez-González J, Arango C, Castro J, González-Pinto A, Ortuño F, et al. Assessment of the increase in variability when combining volumetric data from different scanners. Hum Brain Mapp. 2009; 30:355-68.

[6] Lodge MA, Lesniak W, Gorin MA, Pienta KJ, Rowe SP, Pomper MG. Measurement of PET Quantitative Bias In Vivo. J Nucl Med. 2021; 62:732-7.

[7] Beyer T, Antoch G, Müller S, Egelhof T, Freudenberg LS, Debatin J, et al. Acquisition protocol considerations for combined PET/CT imaging. J Nucl Med. 2004; 45:25S-35S.

[8] Boellaard R, Oyen WJG, Hoekstra CJ, Hoekstra OS, Visser EP, Willemsen AT, et al. The Netherlands protocol for standardisation and quantification of FDG whole body PET studies in multi-centre trials. Eur J Nucl Med Mol Imaging. 2008; 35:2320-33.

[9] Doot RK, Scheuermann JS, Christian PE, Karp JS, Kinahan PE. Instrumentation factors affecting variance and bias of quantifying tracer uptake with PET/CT. Med Phys. 2010; 37:6035-46.

[10] Snaith B, Field L, Lewis EF, Flintham K. Variation in pelvic radiography practice: Why can we not standardise image acquisition techniques? Radiography (Lond). 2019; 25:374-7.

[11] Polesello GC, Nakao TS, de Queiroz MC, Daniachi D, Ricioli W Jr, Guimarães RP, et al. PROPOSAL FOR STANDARDIZATION OF RADIOGRAPHIC STUDIES ON THE HIP AND PELVIS. Rev Bras Ortop. 2015; 46:634-42.

[12] An G, Hong L, Zhou XB, Yang Q, Li MQ, Tang XY. Accuracy and efficiency of computer-aided anatomical analysis using 3D visualization software based on semi-automated and automated segmentations. Ann Anat. 2017; 210:76-83.

[13] Khan U, Yasin A, Abid M, Shafi I, Khan SA. A Methodological Review of 3D Reconstruction Techniques in Tomographic Imaging. J Med Syst. 2018; 42:190.

[14] Kim K, Habas PA, Rajagopalan V, Scott JA, Corbett-Detig JM, Rousseau F, et al. Bias field inconsistency correction of motion-scattered multislice MRI for improved 3D image reconstruction. IEEE Trans Med Imaging. 2011; 30:1704-12.

[15] Carapella V, Puchta H, Lukaschuk E, Marini C, Werys K, Neubauer S, et al. Standardized image post-processing of cardiovascular magnetic resonance T1-mapping reduces variability and improves accuracy and consistency in myocardial tissue characterization. Int J Cardiol. 2020; 298:128-34.

[16] Al-Hayek Y, Zheng X, Hayre C, Spuur K. The influence of patient positioning on radiation dose in CT imaging: A narrative review. J Med Imaging Radiat Sci. 2022; 53:737-47.

[17] Lange T, Backhaus SJ, Schulz A, Evertz R, Schneider P, Kowallick JT, et al. Inter-study reproducibility of cardiovascular magnetic resonance-derived hemodynamic force assessments. Sci Rep. 2024; 14:634.

[18] Ismailov T, Khamitova Z, Jumadilova D, Khissamutdinov N, Toktarbay B, Zholshybek N, et al. Reliability of left ventricular hemodynamic forces derived from feature-tracking cardiac magnetic resonance. PLoS One. 2024; 19:e0306481.

[19] Basics of Image Processing. In: Alberich-Bayarri Á, Bellvís-Bataller F, editors. In: The Facts and Challenges of Data Harmonization to Improve Radiomics Reproducibility. Springer Nature; 2024.

[20] Kaalep A, Sera T, Rijnsdorp S, Yaqub M, Talsma A, Lodge MA, et al. Feasibility of state of the art PET/CT systems performance harmonisation. Eur J Nucl Med Mol Imaging. 2018; 45:1344-61.

[21] Panetta JV, Daube-Witherspoon ME, Karp JS. Validation of phantom-based harmonization for patient harmonization. Med Phys. 2017;44:3534-44.

[22] Gatidis S, Kart T, Fischer M, Winzeck S, Glocker B, Bai W, et al. Better Together: Data Harmonization and Cross-Study Analysis of Abdominal MRI Data From UK Biobank and the German National Cohort. Invest Radiol. 2023; 58:346-54.

[23] Kraff O, May MW. Multi-center QA of ultrahigh-field systems. MAGMA. 2025; 38:519-32.

[24] Seoni S, Shahini A, Meiburger KM, Marzola F, Rotunno G, Acharya UR, et al. All you need is data preparation: A systematic review of image harmonization techniques in Multi-center/device studies for medical support systems. Comput Methods Programs Biomed. 2024; 250:108200.

[25] Mali SA, Ibrahim A, Woodruff HC, Andrearczyk V, Müller H, Primakov S, et al. Making Radiomics More Reproducible across Scanner and Imaging Protocol Variations: A Review of Harmonization Methods. J Pers Med. 2021; 11:842.

[26] Whybra P, Zwanenburg A, Andrearczyk V, Schaer R, Apte AP, Ayotte A, et al. The image biomarker standardization initiative: standardized convolutional filters for reproducible radiomics and enhanced clinical insights. Radiology. 2024; 310:e231319.

[27] Meyer M, Ronald J, Vernuccio F, Nelson RC, Ramirez-Giraldo JC, Solomon J, et al. Reproducibility of CT Radiomic Features within the Same Patient: Influence of Radiation Dose and CT Reconstruction Settings. Radiology. 2019; 293:583-91.

[28] Orlhac F, Frouin F, Nioche C, Ayache N, Buvat I. Validation of A Method to Compensate Multicenter Effects Affecting CT Radiomics. Radiology. 2019; 291:53-9.

[29] Horng H, Singh A, Yousefi B, Cohen EA, Haghighi B, Katz S, et al. Generalized ComBat harmonization methods for radiomic features with multi-modal distributions and multiple batch effects. Sci Rep. 2022; 12:4493.

[30] Jodoin PM, Edde M, Girard G, Dumais F, Theaud G, Dumont M, et al. Challenges and best practices when using ComBAT to harmonize diffusion MRI data. Sci Rep. 2025; 15:41508.

[31] Salustri A, Tonti G, Zhankorazova A, Zholshybek N, Toktarbay B, Khamitova Z, et al. Left ventricular hemodynamic forces: gaining insight into left ventricular function. Explor Cardiol. 2025; 3:101257.

[32] Cheng C, Messerschmidt L, Bravo I, Waldbauer M, Bhavikatti R, Schenk C, et al. A General Primer for Data Harmonization. Sci Data. 2024; 11:152.

[33] Papadimitroulas P, Brocki L, Chung NC, Marchadour W, Vermet F, Gaubert L, et al. Artificial intelligence: Deep learning in oncological radiomics and challenges of interpretability and data harmonization. Phys Med. 2021; 83:108-21.

[34] Bashyam VM, Doshi J, Erus G, Srinivasan D, Abdulkadir A, Singh A, et al. Deep Generative Medical Image Harmonization for Improving Cross-Site Generalization in Deep Learning Predictors. J Magn Reson Imaging. 2022; 55:908-16.

[35] Bento M, Fantini I, Park J, Rittner L, Frayne R. Deep Learning in Large and Multi-Site Structural Brain MR Imaging Datasets. Front Neuroinform. 2022; 15:805669.