Abstract
Fetal movement, particularly prominent during second-trimester ultrasound (US) examinations poses a major challenge for motion-aware applications such as 3D reconstruction, standard-plane acquisition, and robotic US navigation. Reliable quantification of fetal motion is therefore essential. However, automated fetal motion analysis remains fundamentally ill-posed because frame-to-frame image changes arise from two independent and entangled sources: probe motion and fetal motion. While probe motion can be measured using external tracking systems, fetal pose cannot be directly observed during routine scanning, resulting in a lack of datasets with explicit fetal-motion supervision.
To address this gap, we present FLAME-US, a data-engineering framework that generates explicit fetal-motion supervision by simulating independent fetal and probe motions within patient-derived 3D fetal head volumes, producing trajectories with known fetal and probe Special Euclidean Group (SE(3)) poses. We further introduce Anatomical Landmark Residuals (ALRs), defined as per-frame landmark displacements attributable exclusively to fetal motion and decoupled from probe movement. Built from 25 patient-derived fetal head volumes, FLAME-US generates an open-source dataset of 10,000 motion trajectories annotated with fetal and probe poses and seven ALRs. Validation shows that ALRs strongly correlate with fetal motion magnitude (Pearson correlation coefficient $r=0.74$, $p<0.001$). Baseline experiments demonstrate that the generated annotations are learnable, achieving fetal rotation error of 0.438 radians, an 85.7\% reduction in error from the lowest to highest motion-magnitude quintile. By providing explicit fetal-motion supervision where direct measurement is not possible, FLAME-US enables future learning-based approaches for fetal motion quantification and motion-aware ultrasound systems
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/DEMI_016.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=hvIf9W4vZL
BibTex
@InProceedings{BalSow_FLAMEUS_MICCAISAT2026,
author = { Balaji, Sowjanya AND A., Anusha AND Ram, Keerthi AND Ayyasamy, Shyam AND Lakshmanan, Manojkumar AND Sivaprakasam, Mohanasankar},
title = { { FLAME-US: A Landmark-Guided Dataset for Quantitative Fetal Motion Analysis in Freehand Ultrasound } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17276},
month = {pending},
page = {pending}
}