Cuadra, M. B. | Search bioRxiv

bioRxiv2024

Maturation-informed synthetic Magnetic Resonance Images of the Developing Human Fetal Brain

Magnetic resonance imaging is a powerful modality to investigate abnormal developmental patterns in utero. However, since it is not the first-line diagnostic tool in this sensitive population, data remain scarce and heterogeneous between scanners and centers. In addressing the data scarcity issue while generating data representative of real fetal brain MRI, we proposed FaBiAN, a Fetal Brain magnetic resonance Acquisition Numerical phantom. Here, we present a novel synthetic dataset of 594 two-dimensional, low-resolution series of T2-weighted images corresponding to 78 developing human fetal brains between 20.0 and 34.8 weeks of gestational age. Data are generated with substantive improvements from the original FaBiAN to account for local heterogeneities within white matter tissues throughout maturation. These synthetic-yet-highly-realistic images cover both healthy and pathological development trajectories simulated with standard clinical settings and anatomically informed by the Fetal Tissue Annotations (FeTA) dataset. Two independent radiologists qualitatively assessed the realism of the simulated images. We also quantitatively demonstrate the simulated datas increased fidelity to real data compared to the previous FaBiAN version. The reuse potential of the proposed dataset was also evaluated in the context of automated fetal brain tissue segmentation. Besides, our dataset that combines images generated from various clinical scenarios has been made publicly available to support the continuous endeavor of the community to develop advanced post-processing methods aswell as cutting-edge artificial intelligence models.

bioinformatics↗

bioRxiv2021

Through-plane super-resolution with autoencoders in diffusion magnetic resonance imaging of the developing human brain

Fetal brain diffusion magnetic resonance images are often acquired with a lower through-plane than in-plane resolution. This anisotropy is often overcome by classical upsampling methods such as linear or cubic interpolation. In this work, we employ an unsupervised learning algorithm using an autoencoder neural network to enhance the through-plane resolution by leveraging a large amount of data. Our framework, which can also be used for slice outliers replacement, overperformed conventional interpolations quantitatively and qualitatively on pre-term newborns of the developing Human Connectome Project. The evaluation was performed on both the original diffusion-weighted signal and on the estimated diffusion tensor maps. A byproduct of our autoencoder was its ability to act as a denoiser. The network was able to generalize to fetal data with different levels of motion and we qualitatively showed its consistency, hence supporting the relevance of pre-term datasets to improve the processing of fetal brain images.

bioinformatics↗