Deep cascaded registration and weakly-supervised segmentation of fetal brain MRI

Deformable image registration is a cornerstone of many medical image analysis applications, particularly in the context of fetal brain magnetic resonance imaging (MRI), where precise registration is essential for studying the rapidly evolving fetal brain during pregnancy and potentially identifying neurodevelopmental abnormalities. While deep learning has become the leading approach for medical image registration, traditional convolutional neural networks (CNNs) often fall short in capturing fine image details due to their bias toward low spatial frequencies. To address this challenge, we introduce a deep learning registration framework comprising multiple cascaded convolutional networks. These networks predict a series of incremental deformation fields that transform the moving image at various spatial frequencies, ensuring accurate alignment with the fixed image. This multi-resolution approach allows for a more accurate and detailed registration process, capturing both coarse and fine image structures.We demonstrate the superior performance of our method by a significant margin compared to other state-of-the-art methods, including other multi-resolution techniques. Our method outperforms existing state-of-the-art techniques, including other multi-resolution strategies, by a substantial margin. Additionally, we integrate this cascaded registration framework into a multi-atlas segmentation pipeline, demonstrating competitive performance against the nnU-Net while using a significantly smaller set of annotated images as atlases. This approach is particularly valuable in the context of fetal brain MRI, where annotated datasets are limited. Our pipeline for registration and multi-subject segmentation is publicly available at \url{https://github.com/ValBcn/CasReg}.

COMTE Valentin;  ALENYA Mireia;  URRU Andrea;  RECOBER Judith;  NAKAKI Ayako;  CROVETTO Francesca;  CAMARA REY Oscar;  GRATACÓS Eduard;  EIXARCH Elisenda;  CRISPI Fàtima;  PIELLA Gemma;  CERESA Mario;  GONZALEZ BALLESTER Miguel Angel; 

2026-03-06

CELL PRESS

JRC134476

2405-8440 (online),   

https://www.sciencedirect.com/science/article/pii/S2405844024161798,    https://publications.jrc.ec.europa.eu/repository/handle/JRC134476,   

10.1016/j.heliyon.2024.e40148 (online),