IGUANe: a 3D generalizable CycleGAN for multicenter harmonization of brain MR images

TL;DR

Comparisons with other harmonization and normalization methods suggest that IGUANe better preserves individual information in MR images and is more suitable for maintaining and reinforcing variabilities related to age and AD.

Abstract

In MRI studies, the aggregation of imaging data from multiple acquisition sites enhances sample size but may introduce site-related variabilities that hinder consistency in subsequent analyses. Deep learning methods for image translation have emerged as a solution for harmonizing MR images across sites. In this study, we introduce IGUANe (Image Generation with Unified Adversarial Networks), an original 3D model that leverages the strengths of domain translation and straightforward application of style transfer methods for multicenter brain MR image harmonization. IGUANe extends CycleGAN by integrating an arbitrary number of domains for training through a many-to-one architecture. The framework based on domain pairs enables the implementation of sampling strategies that prevent confusion between site-related and biological variabilities. During inference, the model can be applied to any image, even from an unknown acquisition site, making it a universal generator for harmonization. Trained on a dataset comprising T1-weighted images from 11 different scanners, IGUANe was evaluated on data from unseen sites. The assessments included the transformation of MR images with traveling subjects, the preservation of pairwise distances between MR images within domains, the evolution of volumetric patterns related to age and Alzheimer$'$s disease (AD), and the performance in age regression and patient classification tasks. Comparisons with other harmonization and normalization methods suggest that IGUANe better preserves individual information in MR images and is more suitable for maintaining and reinforcing variabilities related to age and AD. Future studies may further assess IGUANe in other multicenter contexts, either using the same model or retraining it for applications to different image modalities. IGUANe is available at https://github.com/RocaVincent/iguane_harmonization.git.

Figures (14)

• Figure 1: Representation of the modules in the IGUANe framework.
• Figure 1: Probability distributions of age and sampling for IGUANe training in the Training dataset. Each subfigure indicates the age distributions of SALD and of a source site as well as the sampling probabilities used for training. The graduations on the x-axes indicate the age ranges used for bias sampling. The color intensities indicate the proportion for each age range.
• Figure 1: Visualization of MR images harmonized with the different methods implemented. One image from the reference domain (SALD) and one from each study in the Generalization dataset were randomly sampled, with a middle axial slice shown for each. preproc refers to the images obtained after preprocessing for the corresponding harmonization approach.
• Figure 1: Inter-image Euclidean distances in a sample of 10 images from the SALD dataset before and after IGUANe harmonization. The 45 points correspond to the 45 image pairs. The distances are expressed in tens of thousands.
• Figure 1: Correlation between age and gray-matter (GM) volume in the Generalization dataset for the ablation studies. The X and Y axes correspond to ages and GM volumes (divided by the total intracranial volume), respectively. The linear least-squares regression line is plotted on each subfigure.