Get Started
Home
Topics
Search
Library
Research questionHow can automated aortic segmentation in 4D flow MRI remain time-resolved without dense annotations or excessive computation?Reproducible hemodynamic measurements require accurate segmentation throughout the cardiac cycle. However, dense 4D labels are scarce, while processing volumetric time-series data is computationally demanding.
Computer Vision
Health
Image & Video Processing
Machine Learning
Research Paper
Latest papersRecent research connected to this question, newest first.Segmentation of the aorta in 4D flow MRI using 4D convolutional kernels and learning from sparse annotationsThe evidence covers a fully automated 4D U-Net trained on 268 scans from eight centers and two vendors using sparse 4D labels derived from existing 2D expert contours and centerlines. Evaluation included 32 internal test scans and 30 independent post-contrast scans from a different site, protocol, and annotator, with comparisons against frame-wise 3D networks and semi-automatic references. Reported outcomes include agreement with time-resolved annotations and expert-derived hemodynamic measurements; the model is publicly available.research paper · Sep 3, 2026
Related questions
How can low-resolution, noisy 4D flow MRI yield accurate carotid hemodynamic biomarkers?How can brain MRI super-resolution preserve continuous tissue mixtures at partial-volume tissue transitions?How can we reliably segment multiple pediatric brain structures in ultra-low-field MRI when annotations disagree?How can longitudinal MRI models forecast neurodegenerative progression without confusing stable anatomy or noise with disease change?