Abstract:We develop a novel harmonization approach for T1‐weighted magnetic resonance imaging using a style‐encoding generative adversarial network that can be used to harmonize entire images for a variety of international, multi‐cohort, neuroimaging collaborations. Results demonstrated that this model avoids the need to control for clinical or demographic information. We showed that our harmonization removed the cross‐site variances, while preserving the anatomical information and clinical meaningful patterns. Recent work within neuroimaging consortia have aimed to identify reproducible, and often subtle, brain signatures of psychiatric or neurological conditions. To allow for high‐powered brain imaging analyses, it is often necessary to pool MR images that were acquired with different protocols across multiple scanners. Current retrospective harmonization techniques have shown promise in removing site‐related image variation. However, most statistical approaches may over‐correct for technical, scanning‐related, variation as they cannot distinguish between confounded image‐acquisition based variability and site‐related population variability. Such statistical methods often require that datasets contain subjects or patient groups with similar clinical or demographic information to isolate the acquisition‐based variability. To overcome this limitation, we consider site‐related magnetic resonance (MR) imaging harmonization as a style transfer problem rather than a domain transfer problem. Using a fully unsupervised deep‐learning framework based on a generative adversarial network (GAN), we show that MR images can be harmonized by inserting the style information encoded from a single reference image, without knowing their site/scanner labels a priori. We trained our model using data from five large‐scale multisite datasets with varied demographics. Results demonstrated that our style‐encoding model can harmonize MR images, and match intensity profiles, without relying on traveling subjects. This model also avoids the need to control for clinical, diagnostic, or demographic information. We highlight the effectiveness of our method for clinical research by comparing extracted cortical and subcortical features, brain‐age estimates, and case–control effect sizes before and after the harmonization. We showed that our harmonization removed the site‐related variances, while preserving the anatomical information and clinical meaningful patterns. We further demonstrated that with a diverse training set, our method successfully harmonized MR images collected from unseen scanners and protocols, suggesting a promising tool for ongoing collaborative studies. Source code is released in USC‐IGC/style_transfer_harmonization (github.com).

Harmonization techniques for machine learning studies using multi-site functional MRI data

A deep learning-based multisite neuroimage harmonization framework established with a traveling-subject dataset

Harmonization of Diffusion MRI-based Structural Connectomes for Multi-site Studies

Comprehensive evaluation of harmonization on functional brain imaging for multisite data-fusion

Addressing multi-site functional MRI heterogeneity through dual-expert collaborative learning for brain disease identification

Harmonization of Multi-Site Functional MRI Data with Dual-Projection Based ICA Model

Harmonization of resting-state functional MRI data across multiple imaging sites via the separation of site differences into sampling bias and measurement bias

Enhancing Autism Spectrum Disorder identification in multi-site MRI imaging: A multi-head cross-attention and multi-context approach for addressing variability in un-harmonized data

Site effects removing and signal enhancement using dual-projection based ICA model

Multi-Site Harmonization of Diffusion MRI Data Via Method of Moments

Efficacy of MRI data harmonization in the age of machine learning: a multicenter study across 36 datasets

Efficacy of MRI data harmonization in the age of machine learning. A multicenter study across 36 datasets

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

Multi-site MRI Harmonization Via Attention-Guided Deep Domain Adaptation for Brain Disorder Identification.

Cross-site harmonization of diffusion MRI data without matched training subjects

Style transfer generative adversarial networks to harmonize multisite MRI to a single reference image to avoid overcorrection

Removal of Site Effects and Enhancement of Signal Using Dual Projection Independent Component Analysis for Pooling Multi-Site MRI Data.

Learning Multi-Site Harmonization of Magnetic Resonance Images Without Traveling Human Phantoms

Harmonization Benchmarking Tool for Neuroimaging Datasets

Intensity warping for multisite MRI harmonization

Unpaired Volumetric Harmonization of Brain MRI with Conditional Latent Diffusion