Published July 12, 2026 | Version v1
Dataset Open

FeEcho4D: A Benchmark Dataset for 4D Reconstruction of the Fetal Left Ventricle from Echocardiography

Description

FeEcho4D is the first publicly available benchmark dataset for 4D (3D+t) fetal echocardiography, developed to advance research in cardiac segmentation, motion analysis, and 4D reconstruction. The dataset comprises 52 annotated fetal Spatio-Temporal Image Correlation (STIC) studies, containing 1,845 annotated 3D volumes (approximately 68,265 radial slices) with temporally coherent left ventricular (LV) segmentations and 3D meshes. To encourage the development of semi-supervised and self-supervised learning methods, the dataset also includes 50 additional unannotated 4D STIC studies. FeEcho4D introduces a novel radial slicing representation that produces consistent U-shaped anatomical views, enabling more reliable manual annotation and facilitating geometry-aware deep learning. The dataset serves as a standardized benchmark for developing and evaluating AI methods for fetal cardiac segmentation, motion estimation, 3D/4D reconstruction, shape modelling, and functional biomarker estimation in fetal cardiac ultrasound. Details in https://feecho4d.github.io/Website/.
Patient0xx/
├── image/
│   ├── Patient0xx_slice0yytime0zz.png
│   ├── ...
│   └── Radially extracted 2D fetal echocardiography slices
│
├── mask/
│   ├── Patient0xx_slice0yytime0zz.png
│   ├── ...
│   └── Manual LV/MYO segmentations
│       ├── LV cavity   = 127
│       └── MYO         = 255
│
├── mesh/
│   ├── mesh_time0zz.obj
│   ├── ...
│   └── Reconstructed 3D LV meshes over time
│
├── scribble/
│   ├── Patient0xx_slice0yytime0zz_scribbleTypeN.png
│   ├── ...
│   └── Weak scribble prompts (N = 1 ... 10)
│
├── Patient0xx_bbox.csv
│   └── Bounding-box prompts:
│       ├── filename
│       ├── xmin
│       ├── ymin
│       ├── xmax
│       └── ymax
│
├── Patient0xx.mp4
│   └── 3D LV mesh visualization over time
│
└── Patient0xx_info.cfg
    ├── diagnosis = healthy / ToF / dTGA
    ├── x_mm
    ├── y_mm
    ├── z_mm
    ├── num_frames
    ├── ED_time
    └── ES_time
 

Files

FeEcho4D.zip

Files (11.4 GB)

Name Size
md5:b9c07a261849b224a743c0ec2037db6c
11.4 GB Preview Download

Additional details

Additional titles

Alternative title
4D Reconstruction of Fetal Left Ventricle from Echocardiography via 2.5D Radial Segmentation and Graph-Fourier Reconstruction

Identifiers

DOI
10.1109/TMI.2026.3707322
URL
https://feecho4d.github.io/Website/

Related works

Is described by
Publication: 10.1109/TMI.2026.3707322 (DOI)
Is part of
Publication: 10.1016/j.bspc.2026.110191 (DOI)

Software

Repository URL
https://github.com/kamruleee51/FeEcho4D