Synthetic magnetic resonance images for domain adaptation: Application to fetal brain tissue segmentation

TL;DR

This paper introduces a method using synthetic fetal brain MRI images generated by FaBiAN to improve the domain adaptation of deep learning segmentation models, significantly enhancing accuracy across multiple brain tissues.

Contribution

The study presents a novel approach of using synthetic annotated images for domain adaptation in fetal brain segmentation, addressing data scarcity and heterogeneity issues.

Findings

Segmentation accuracy improved across multiple brain tissues.

Synthetic data effectively enhances domain adaptation.

Method reduces need for extensive real annotated datasets.

Abstract

The quantitative assessment of the developing human brain in utero is crucial to fully understand neurodevelopment. Thus, automated multi-tissue fetal brain segmentation algorithms are being developed, which in turn require annotated data to be trained. However, the available annotated fetal brain datasets are limited in number and heterogeneity, hampering domain adaptation strategies for robust segmentation. In this context, we use FaBiAN, a Fetal Brain magnetic resonance Acquisition Numerical phantom, to simulate various realistic magnetic resonance images of the fetal brain along with its class labels. We demonstrate that these multiple synthetic annotated data, generated at no cost and further reconstructed using the target super-resolution technique, can be successfully used for domain adaptation of a deep learning method that segments seven brain tissues. Overall, the accuracy of the segmentation is significantly enhanced, especially in the cortical gray matter, the white matter, the cerebellum, the deep gray matter and the brain stem.