Towards contrast- and pathology-agnostic clinical fetal brain MRI segmentation using SynthSeg
Year: 2026
Type: article
Abstract: Magnetic resonance imaging (MRI) has played a crucial role in fetal neurodevelopmental research. Structural annotations of MR images are an important step for quantitative analysis of the developing human brain, with Deep Learning providing an automated alternative for this otherwise tedious manual process. However, segmentation performances of Convolutional Neural Networks often suffer from domain shift, where the network fails when applied to subjects that deviate from the distribution with which it is trained on. In this work, we aim to train networks capable of automatically segmenting fetal brain MRIs with a wide range of domain shifts pertaining to differences in subject physiology and acquisition environments, in particular shape-based differences commonly observed in pathological cases. We introduce a novel data-driven train-time sampling strategy that seeks to fully exploit the diversity of a given training dataset to enhance the domain generalizability of the trained networks. We adapted our sampler, together with other existing data augmentation techniques, to the SynthSeg framework, a generator that utilizes domain randomization to generate diverse training data. We ran thorough experimentations and ablation studies on a wide range of training/testing data to test the validity of the approaches. Our networks achieved notable improvements in the segmentation quality on testing subjects with intense anatomical abnormalities (p < 1e-4), though at the cost of a slighter decrease in performance in cases with fewer abnormalities. Our work also lays the foundation for future works on creating and adapting data-driven sampling strategies for other training pipelines.
Source: NeuroImage
Institutions: ETH Zurich, University Children's Hospital Zurich, University of Zurich, Consejo Superior de Investigaciones Científicas, Instituto de Instrumentación para Imagen Molecular, Universitat Politècnica de València, Medical University of Vienna, King's College London, Friedrich-Alexander-Universität Erlangen-Nürnberg, Universitätsklinikum Erlangen, Centre Hospitalier Universitaire Vaudois, Centre d'Imagerie BioMedicale, École Polytechnique Fédérale de Lausanne, University of Lausanne
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $3,540
Funders: Anna Mueller Grocholski-Stiftung, National Science Foundation, UK Research and Innovation, European Commission, Deutsche Forschungsgemeinschaft, École Polytechnique Fédérale de Lausanne, Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung, Université de Genève, Université de Lausanne, Centre d'Imagerie BioMédicale, Centre Hospitalier Universitaire Vaudois, Universität Zürich, Vontobel-Stiftung, Seventh Framework Programme, European Research Council, H2020 European Research Council, FP7 Ideas: European Research Council, HORIZON EUROPE European Research Council
Awards: EARTHWORM: pEristAlsis in Real-Time Human mri to study the interWOven fRequency & Microstructural properties, G192932977, Brain age and digital twins as markers of infant neurodevelopment: a machine-learning approach using multinational MRI data, G3207356259, G33188248, LONG-SB Study: Revealing the neural basis of long-term cognitive development of children with open spina bifida, G5537678747, Tackling domain shifts in pediatric neuroimaging: bridging advanced computational MR techniques and clinical practice, G5762252440, Advanced super-resolution reconstruction methods for quantitative magnetic resonance imaging of the developing fetal brain, G6027190877, Intelligente MR Bildgebung, G7553243680, Multicentric study of Fetal Abnormal Cortical Trajectory with standardised and privacy-preserving method on fetal MRI, G878044590, G8792368603, G896728979, The Developing Human Connectome Project
OpenAlex ID: W4415160276