Statistical harmonization corrects site effects in functional connectivity measurements from multi‐site fMRI data

Abstract

Acquiring resting-state functional magnetic resonance imaging (fMRI) datasets at multiple MRI scanners and clinical sites can improve statistical power and generalizability of results. However, multi-site neuroimaging studies have reported considerable nonbiological variability in fMRI measurements due to different scanner manufacturers and acquisition protocols. These undesirable sources of variability may limit power to detect effects of interest and may even result in erroneous findings. Until now, there has not been an approach that removes unwanted site effects. In this study, using a relatively large multi-site (4 sites) fMRI dataset, we investigated the impact of site effects on functional connectivity and network measures estimated by widely used connectivity metrics and brain parcellations. The protocols and image acquisition of the dataset used in this study had been homogenized using identical MRI phantom acquisitions from each of the neuroimaging sites; however, intersite acquisition effects were not completely eliminated. Indeed, in this study, we found that the magnitude of site effects depended on the choice of connectivity metric and brain atlas. Therefore, to further remove site effects, we applied ComBat, a harmonization technique previously shown to eliminate site effects in multi-site diffusion tensor imaging (DTI) and cortical thickness studies. In the current work, ComBat successfully removed site effects identified in connectivity and network measures and increased the power to detect age associations when using optimal combinations of connectivity metrics and brain atlases. Our proposed ComBat harmonization approach for fMRI-derived connectivity measures facilitates reliable and efficient analysis of retrospective and prospective multi-site fMRI neuroimaging studies.

Nonreplication of functional connectivity differences in autism spectrum disorder across multiple sites and denoising strategies

“…Fortunately, these factors, while they do contribute to across-site variance, tend to be small in terms of effect size (Brown et al, 2011;Dansereau et al, 2017;Noble et al, 2017) or result in localized differences (Nair et al, 2018), consistent with our finding that group-averaged connectomes were highly reliable across sites. To further increase chances of replication, either a priori coordination and standardization of procedures (Glover et al, 2012) or the implementation of postprocessing methods designed to increase multisite data harmonization would both be possibilities (Yamashita et al, 2019;Yu et al, 2018).…”

“…Further complicating the picture is that site effects, or variation across different scanning sites, have been reported in several studies of both task-based and resting-state fMRI (Brown et al, 2011;Dansereau et al, 2017;Noble et al, 2017;Turner et al, 2013;Yamashita et al, 2019;Yan, Craddock, Zuo, Zang, & Milham, 2013;Yu et al, 2018). Different sites present many potential sources of variation, including differences in participant (i.e., cohort) characteristics, image acquisition parameters, scanners, scan procedures, and more.…”

Intensity warping for multisite MRI harmonization

“…Unlike intensity normalization methods, which target intensity unit effects, harmonization methods aim to reduce scanner effects so that downstream analyses are more comparable across sites and scanners (Fortin et al, 2017;Yu et al, 2018). Fortin et al (2018) described a voxel-wise regression method, based on tools from genomics, that harmonizes cortical thickness measurements from MRI scans.…”

“…This method succeeds in removing scanner effects for measurements extracted from each image; in contrast, our goal in the present study was to develop an effective harmonization method that can be applied to the entire brain. Similar tools from genomics are used to correct for scanner effects in multisite diffusion tensor imaging data (Fortin et al, 2017) and multisite functional MRI data (Yu et al, 2018). However, these harmonization methods require spatial registration to a population template, which can lower image resolution and make it challenging to detect important disease features such as MS lesions.…”

Harmonization of large multi-site imaging datasets: Application to 10,232 MRIs for the analysis of imaging patterns of structural brain change throughout the lifespan

“…Other communities that handle high dimensional data-integration across multiple sites have faced the necessity of harmonization. Among the available methods, ComBat, which was originally proposed to remove batch effects in genomics data (Johnson et al, 2007), has been recently adapted to diffusion tensor imaging data (Fortin et al, 2017), cortical thickness measurements (Fortin et al, 2018), and functional connectivity matrices (Yu et al, 2018). The method was shown to remove unwanted sources of variability, specifically site differences, while preserving variations due to other biologically-relevant covariates in the data.…”