Abstract
Four-chamber segmentation and left ventricular ejection fraction (LVEF) calculation in four-dimensional (4D) echocardiography are vital for comprehensive cardiac assessment. This paper presents the first study on fully automated four-chamber segmentation in 4D echocardiography. We introduce a two-stage segmentation strategy to improve end-diastolic and end-systolic frame segmentation, directly enhancing LVEF accuracy. The novel dihedral and convex loss functions are proposed to enforce inter-chamber spatial relationships and irregular chamber shapes. A new 4D echocardiography dataset comprising 29 sequences with 145 annotated frames is established. Using 3D-UXNet as the backbone network, our method achieves a mean Dice for the four chambers of 0.824, a mean Hausdorff Distance at the 95th percentile in voxels of 7.036, and a low mean absolute error for LVEF of 0.041 (vs. Annotation) and 0.046 (vs. cardiac magnetic resonance).
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/TIA_016.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: Not Available
BibTex
@InProceedings{WanZhe_3D4CSegNet_MICCAISAT2026,
author = { Wang, Zhenkuan AND Song, Wenming AND Liu, Ting AND Zhang, Wenxuan AND Chen, Kang AND Guo, Zhiling AND Wei, Hongbin AND Han, Min AND Cong, Longfei AND Lin, Muqing AND Xu, Yongli AND An, Xing},
title = { { 3D-4C-SegNet: Four-Chamber Segmentation and LVEF Calculation in 4D Echocardiography } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17269},
month = {pending},
page = {pending}
}