Ultra-ECP: Ellipse-Constrained and Point-Robust Foundation Model Adaptation for Fetal Cardiac Ultrasound Segmentation

Minh H. N. Le, Khanh T. Q. Le, Tuan Vinh, Thanh-Huy Nguyen, Han H. Huynh, Khoa D. Pham, Anh Mai Vu, Hien Q. Kha, Phat K. Nguyen, Ulas Bagci, Min Xu, Carl Yang, Phat K. Huynh, Nguyen Quoc Khanh Le
Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, PMLR 315:2311-2320, 2026.

Abstract

Accurate fetal cardiac segmentation from four-chamber ultrasound images is essential for reliable prenatal biometrics, yet foundation models such as SAM remain sensitive to point-prompt placement, produce anatomically inconsistent masks, and require costly full-model fine-tuning. We introduce Ultra-ECP, a parameter-efficient framework that adapts UltraSAM for robust single-point fetal cardiac segmentation. Ultra-ECP integrates three components: (i) a LoRA-based adaptation applied to the prompt encoder and mask decoder, reducing trainable parameters by over 98%; (ii) an Ellipse-Aware Loss that regularizes predictions toward anatomically plausible elliptical cardiac shapes; and (iii) a Point-Robust Augmentation strategy that simulates click imprecision to enhance robustness. Evaluated on the FOCUS dataset, Ultra-ECP outperforms SAM, MedSAM, and fine-tuned U-Net baselines. For thoracic segmentation, it achieves a mean DSC of 95.09% and HD95 of 25.96 px. For cardiac segmentation, Ultra-ECP obtains a mean DSC of 92.60% and HD95 of 18.25 px, while maintaining stability under point displacements of up to 10 pixels. Predictions are consistently smooth and elliptical, addressing common failure modes of existing approaches. Ultra-ECP provides an effective and computationally lightweight pathway for adapting large vision models to fetal cardiac biometrics, enabling reliable and clinically practical semi-automated tools.

Cite this Paper

BibTeX
@InProceedings{pmlr-v315-le26a, title = {Ultra-ECP: Ellipse-Constrained and Point-Robust Foundation Model Adaptation for Fetal Cardiac Ultrasound Segmentation}, author = {Le, Minh H. N. and Le, Khanh T. Q. and Vinh, Tuan and Nguyen, Thanh-Huy and Huynh, Han H. and Pham, Khoa D. and Vu, Anh Mai and Kha, Hien Q. and Nguyen, Phat K. and Bagci, Ulas and Xu, Min and Yang, Carl and Huynh, Phat K. and Le, Nguyen Quoc Khanh}, booktitle = {Proceedings of The 9th International Conference on Medical Imaging with Deep Learning}, pages = {2311--2320}, year = {2026}, editor = {Huo, Yuankai and Gao, Mingchen and Kuo, Chang-Fu and Jin, Yueming and Deng, Ruining}, volume = {315}, series = {Proceedings of Machine Learning Research}, month = {08--10 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v315/main/assets/le26a/le26a.pdf}, url = {https://proceedings.mlr.press/v315/le26a.html}, abstract = {Accurate fetal cardiac segmentation from four-chamber ultrasound images is essential for reliable prenatal biometrics, yet foundation models such as SAM remain sensitive to point-prompt placement, produce anatomically inconsistent masks, and require costly full-model fine-tuning. We introduce Ultra-ECP, a parameter-efficient framework that adapts UltraSAM for robust single-point fetal cardiac segmentation. Ultra-ECP integrates three components: (i) a LoRA-based adaptation applied to the prompt encoder and mask decoder, reducing trainable parameters by over 98%; (ii) an Ellipse-Aware Loss that regularizes predictions toward anatomically plausible elliptical cardiac shapes; and (iii) a Point-Robust Augmentation strategy that simulates click imprecision to enhance robustness. Evaluated on the FOCUS dataset, Ultra-ECP outperforms SAM, MedSAM, and fine-tuned U-Net baselines. For thoracic segmentation, it achieves a mean DSC of 95.09% and HD95 of 25.96 px. For cardiac segmentation, Ultra-ECP obtains a mean DSC of 92.60% and HD95 of 18.25 px, while maintaining stability under point displacements of up to 10 pixels. Predictions are consistently smooth and elliptical, addressing common failure modes of existing approaches. Ultra-ECP provides an effective and computationally lightweight pathway for adapting large vision models to fetal cardiac biometrics, enabling reliable and clinically practical semi-automated tools.} }
Endnote
%0 Conference Paper %T Ultra-ECP: Ellipse-Constrained and Point-Robust Foundation Model Adaptation for Fetal Cardiac Ultrasound Segmentation %A Minh H. N. Le %A Khanh T. Q. Le %A Tuan Vinh %A Thanh-Huy Nguyen %A Han H. Huynh %A Khoa D. Pham %A Anh Mai Vu %A Hien Q. Kha %A Phat K. Nguyen %A Ulas Bagci %A Min Xu %A Carl Yang %A Phat K. Huynh %A Nguyen Quoc Khanh Le %B Proceedings of The 9th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2026 %E Yuankai Huo %E Mingchen Gao %E Chang-Fu Kuo %E Yueming Jin %E Ruining Deng %F pmlr-v315-le26a %I PMLR %P 2311--2320 %U https://proceedings.mlr.press/v315/le26a.html %V 315 %X Accurate fetal cardiac segmentation from four-chamber ultrasound images is essential for reliable prenatal biometrics, yet foundation models such as SAM remain sensitive to point-prompt placement, produce anatomically inconsistent masks, and require costly full-model fine-tuning. We introduce Ultra-ECP, a parameter-efficient framework that adapts UltraSAM for robust single-point fetal cardiac segmentation. Ultra-ECP integrates three components: (i) a LoRA-based adaptation applied to the prompt encoder and mask decoder, reducing trainable parameters by over 98%; (ii) an Ellipse-Aware Loss that regularizes predictions toward anatomically plausible elliptical cardiac shapes; and (iii) a Point-Robust Augmentation strategy that simulates click imprecision to enhance robustness. Evaluated on the FOCUS dataset, Ultra-ECP outperforms SAM, MedSAM, and fine-tuned U-Net baselines. For thoracic segmentation, it achieves a mean DSC of 95.09% and HD95 of 25.96 px. For cardiac segmentation, Ultra-ECP obtains a mean DSC of 92.60% and HD95 of 18.25 px, while maintaining stability under point displacements of up to 10 pixels. Predictions are consistently smooth and elliptical, addressing common failure modes of existing approaches. Ultra-ECP provides an effective and computationally lightweight pathway for adapting large vision models to fetal cardiac biometrics, enabling reliable and clinically practical semi-automated tools.
APA
Le, M.H.N., Le, K.T.Q., Vinh, T., Nguyen, T., Huynh, H.H., Pham, K.D., Vu, A.M., Kha, H.Q., Nguyen, P.K., Bagci, U., Xu, M., Yang, C., Huynh, P.K. & Le, N.Q.K.. (2026). Ultra-ECP: Ellipse-Constrained and Point-Robust Foundation Model Adaptation for Fetal Cardiac Ultrasound Segmentation. Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 315:2311-2320 Available from https://proceedings.mlr.press/v315/le26a.html.

Related Material