Towards contrast- and pathology-agnostic clinical fetal brain MRI segmentation using SynthSeg.
Authors
Affiliations (12)
Affiliations (12)
- Center for MR Research, University Children's Hospital Zurich; Lenggstrasse 30, 8008 Zürich, Switzerland; Department of Computer Science, ETH Zurich; Universitätstr. 6, 8092 Zürich, Switzerland. Electronic address: [email protected].
- Center for MR Research, University Children's Hospital Zurich; Lenggstrasse 30, 8008 Zürich, Switzerland; Faculty of Medicine, University of Zurich; Pestalozzistrasse 3 CH-8032 Zurich, Switzerland. Electronic address: [email protected].
- Center for MR Research, University Children's Hospital Zurich; Lenggstrasse 30, 8008 Zürich, Switzerland. Electronic address: [email protected].
- Center for MR Research, University Children's Hospital Zurich; Lenggstrasse 30, 8008 Zürich, Switzerland; MRILab, Institute for Molecular Imaging and Instrumentation (i3M), Spanish National Research Council (CSIC), Universitat Politècnica de València (UPV); Instituto I3M - Edificio 8B, Acceso N, Planta 1. Camino de Vera s/n - 46022, València, Spain. Electronic address: [email protected].
- Computational Imaging Research Lab (CIR), Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna; Spitalgasse 23, A-1090 Vienna, Austria. Electronic address: [email protected].
- Computational Imaging Research Lab (CIR), Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna; Spitalgasse 23, A-1090 Vienna, Austria. Electronic address: [email protected].
- Research Department of Early Life Imaging, School of Biomedical Engineering and Imaging Sciences, King's College London; Strand, London, WC2R 2LS, United Kingdom; Biomedical Engineering Department, School of Biomedical Engineering and Imaging Sciences, King's College London; Strand, London, WC2R 2LS, United Kingdom. Electronic address: [email protected].
- Research Department of Early Life Imaging, School of Biomedical Engineering and Imaging Sciences, King's College London; Strand, London, WC2R 2LS, United Kingdom; Biomedical Engineering Department, School of Biomedical Engineering and Imaging Sciences, King's College London; Strand, London, WC2R 2LS, United Kingdom; Smart Imaging Lab, Radiological Institute, University Hospital Erlangen; Friedrich-Alexander-Universität Erlangen-Nürnberg, Schlossplatz 4, 91054 Erlangen, Germany. Electronic address: [email protected].
- Department of Quantitative Biomedicine, University of Zurich; Winterthurerstrasse 190, 8057 Zurich, Switzerland. Electronic address: [email protected].
- Computational Imaging Research Lab (CIR), Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna; Spitalgasse 23, A-1090 Vienna, Austria; Division of Neuroradiology and Musculoskeletal Radiology, Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna; Währinger Gürtel 18-20, 1090 Vienna, Austria. Electronic address: [email protected].
- CIBM Center for Biomedical Imaging; EPFL AVP CP CIBM, Station 6, 1015, Lausanne Switzerland; Radiology Department, University of Lausanne and Lausanne University Hospital; Rue de Bugnon 21, CH-1011 Lausanne, Vaud, Switzerland. Electronic address: [email protected].
- Center for MR Research, University Children's Hospital Zurich; Lenggstrasse 30, 8008 Zürich, Switzerland; Faculty of Medicine, University of Zurich; Pestalozzistrasse 3 CH-8032 Zurich, Switzerland. Electronic address: [email protected].
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.