Dataset Information

A comparison of methods to harmonize cortical thickness measurements across scanners and sites.

ABSTRACT: Results of neuroimaging datasets aggregated from multiple sites may be biased by site-specific profiles in participants' demographic and clinical characteristics, as well as MRI acquisition protocols and scanning platforms. We compared the impact of four different harmonization methods on results obtained from analyses of cortical thickness data: (1) linear mixed-effects model (LME) that models site-specific random intercepts (LME_INT), (2) LME that models both site-specific random intercepts and age-related random slopes (LME_INT+SLP), (3) ComBat, and (4) ComBat with a generalized additive model (ComBat-GAM). Our test case for comparing harmonization methods was cortical thickness data aggregated from 29 sites, which included 1,340 cases with posttraumatic stress disorder (PTSD) (6.2-81.8 years old) and 2,057 trauma-exposed controls without PTSD (6.3-85.2 years old). We found that, compared to the other data harmonization methods, data processed with ComBat-GAM was more sensitive to the detection of significant case-control differences (Χ²(3) = 63.704, p < 0.001) as well as case-control differences in age-related cortical thinning (Χ²(3) = 12.082, p = 0.007). Both ComBat and ComBat-GAM outperformed LME methods in detecting sex differences (Χ²(3) = 9.114, p = 0.028) in regional cortical thickness. ComBat-GAM also led to stronger estimates of age-related declines in cortical thickness (corrected p-values < 0.001), stronger estimates of case-related cortical thickness reduction (corrected p-values < 0.001), weaker estimates of age-related declines in cortical thickness in cases than controls (corrected p-values < 0.001), stronger estimates of cortical thickness reduction in females than males (corrected p-values < 0.001), and stronger estimates of cortical thickness reduction in females relative to males in cases than controls (corrected p-values < 0.001). Our results support the use of ComBat-GAM to minimize confounds and increase statistical power when harmonizing data with non-linear effects, and the use of either ComBat or ComBat-GAM for harmonizing data with linear effects.

SUBMITTER: Sun D

PROVIDER: S-EPMC9648725 | biostudies-literature | 2022 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A comparison of methods to harmonize cortical thickness measurements across scanners and sites.

Sun Delin D Rakesh Gopalkumar G Haswell Courtney C CC Logue Mark M Baird C Lexi CL O'Leary Erin N EN Cotton Andrew S AS Xie Hong H Tamburrino Marijo M Chen Tian T Dennis Emily L EL Jahanshad Neda N Salminen Lauren E LE Thomopoulos Sophia I SI Rashid Faisal F Ching Christopher R K CRK Koch Saskia B J SBJ Frijling Jessie L JL Nawijn Laura L van Zuiden Mirjam M Zhu Xi X Suarez-Jimenez Benjamin B Sierk Anika A Walter Henrik H Manthey Antje A Stevens Jennifer S JS Fani Negar N van Rooij Sanne J H SJH Stein Murray M Bomyea Jessica J Koerte Inga K IK Choi Kyle K van der Werff Steven J A SJA Vermeiren Robert R J M RRJM Herzog Julia J Lebois Lauren A M LAM Baker Justin T JT Olson Elizabeth A EA Straube Thomas T Korgaonkar Mayuresh S MS Andrew Elpiniki E Zhu Ye Y Li Gen G Ipser Jonathan J Hudson Anna R AR Peverill Matthew M Sambrook Kelly K Gordon Evan E Baugh Lee L Forster Gina G Simons Raluca M RM Simons Jeffrey S JS Magnotta Vincent V Maron-Katz Adi A du Plessis Stefan S Disner Seth G SG Davenport Nicholas N Grupe Daniel W DW Nitschke Jack B JB deRoon-Cassini Terri A TA Fitzgerald Jacklynn M JM Krystal John H JH Levy Ifat I Olff Miranda M Veltman Dick J DJ Wang Li L Neria Yuval Y De Bellis Michael D MD Jovanovic Tanja T Daniels Judith K JK Shenton Martha M van de Wee Nic J A NJA Schmahl Christian C Kaufman Milissa L ML Rosso Isabelle M IM Sponheim Scott R SR Hofmann David Bernd DB Bryant Richard A RA Fercho Kelene A KA Stein Dan J DJ Mueller Sven C SC Hosseini Bobak B Phan K Luan KL McLaughlin Katie A KA Davidson Richard J RJ Larson Christine L CL May Geoffrey G Nelson Steven M SM Abdallah Chadi G CG Gomaa Hassaan H Etkin Amit A Seedat Soraya S Harpaz-Rotem Ilan I Liberzon Israel I van Erp Theo G M TGM Quidé Yann Y Wang Xin X Thompson Paul M PM Morey Rajendra A RA

NeuroImage 20220730

Results of neuroimaging datasets aggregated from multiple sites may be biased by site-specific profiles in participants' demographic and clinical characteristics, as well as MRI acquisition protocols and scanning platforms. We compared the impact of four different harmonization methods on results obtained from analyses of cortical thickness data: (1) linear mixed-effects model (LME) that models site-specific random intercepts (LME<sub>INT</sub>), (2) LME that models both site-specific random int ...[more]

PMID: 35917919

Similar Datasets

Project description:The cerebral cortex is a distinctive part of the mammalian nervous system, displaying a spatial variety in cyto-, chemico-, and myelinoarchitecture. As part of a rich history of histological findings, pioneering anatomists von Economo and Koskinas provided detailed mappings on the cellular structure of the human cortex, reporting on quantitative aspects of cytoarchitecture of cortical areas. Current day investigations into the structure of human cortex have embraced technological advances in Magnetic Resonance Imaging (MRI) to assess macroscale thickness and organization of the cortical mantle in vivo. However, direct comparisons between current day MRI estimates and the quantitative measurements of early anatomists have been limited. Here, we report on a simple, but nevertheless important cross-analysis between the histological reports of von Economo and Koskinas on variation in thickness of the cortical mantle and MRI derived measurements of cortical thickness. We translated the von Economo cortical atlas to a subdivision of the commonly used Desikan-Killiany atlas (as part of the FreeSurfer Software package and a commonly used parcellation atlas in studies examining MRI cortical thickness). Next, values of "width of the cortical mantle" as provided by the measurements of von Economo and Koskinas were correlated to cortical thickness measurements derived from high-resolution anatomical MRI T1 data of 200+ subjects of the Human Connectome Project (HCP). Cross-correlation revealed a significant association between group-averaged MRI measurements of cortical thickness and histological recordings (r = 0.54, P < 0.001). Further validating such a correlation, we manually segmented the von Economo parcellation atlas on the standardized Colin27 brain dataset and applied the obtained three-dimensional von Economo segmentation atlas to the T1 data of each of the HCP subjects. Highly consistent with our findings for the mapping to the Desikan-Killiany regions, cross-correlation between in vivo MRI cortical thickness and von Economo histology-derived values of cortical mantle width revealed a strong positive association (r = 0.62, P < 0.001). Linking today's state-of-the-art T1-weighted imaging to early histological examinations our findings indicate that MRI technology is a valid method for in vivo assessment of thickness of human cortex.

Dataset Information

A comparison of methods to harmonize cortical thickness measurements across scanners and sites.

Publications

A comparison of methods to harmonize cortical thickness measurements across scanners and sites.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets