LOTUS: Latent Outpainting Diffusion Model for Three-Dimensional Ultrasound Stitching

Xing Yao, Runxuan Yu, Nick DiSanto, Ehsan Khodapanah Aghdam, Kanyifeechukwu Jane Oguine, Daiwei Lu, Ange Lou, Jiacheng Wang, Dewei Hu, Gabriel A Arenas, Baris Oguz, Alison Marie Pouch, Nadav Schwartz, Brett Byram, Ipek Oguz
Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, PMLR 301:1740-1754, 2026.

Abstract

3D ultrasound (3DUS) stitching can enlarge the field-of-view (FOV) by registering partially overlapping 3DUS images collected from different probe positions. However, standard registration algorithms frequently encounter difficulties with this task, primarily due to the sector-shaped FOV, which often leads to pronounced local minima, thereby obstructing optimization efforts.To address these limitations, we propose LOTUS, a novel Latent Diffusion Model (LDM) specifically designed for 3DUS FOV outpainting. LOTUS innovatively encodes the 3DUS data into a compact latent space and performs outpainting at test time, effectively extending the sector-shaped FOV into a standard rectangular shape. This transformation facilitates a more robust registration by mitigating the issues of local minima associated with the original FOV shape. Experimental results show that LOTUS significantly improves the accuracy of the registration as well as the efficiency of the outpainting process compared to existing models. The code is available at https://github.com/MedICL-VU/LOTUS.

Cite this Paper

BibTeX
@InProceedings{pmlr-v301-yao26a, title = {LOTUS: Latent Outpainting Diffusion Model for Three-Dimensional Ultrasound Stitching}, author = {Yao, Xing and Yu, Runxuan and DiSanto, Nick and Aghdam, Ehsan Khodapanah and Oguine, Kanyifeechukwu Jane and Lu, Daiwei and Lou, Ange and Wang, Jiacheng and Hu, Dewei and Arenas, Gabriel A and Oguz, Baris and Pouch, Alison Marie and Schwartz, Nadav and Byram, Brett and Oguz, Ipek}, booktitle = {Proceedings of The 8th International Conference on Medical Imaging with Deep Learning}, pages = {1740--1754}, year = {2026}, editor = {Tasdizen, Tolga and Elhabian, Shireen and Summers, Ronald and Chen, Chen and Koch, Lisa and Zhuang, Yan}, volume = {301}, series = {Proceedings of Machine Learning Research}, month = {09--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v301/main/assets/yao26a/yao26a.pdf}, url = {https://proceedings.mlr.press/v301/yao26a.html}, abstract = {3D ultrasound (3DUS) stitching can enlarge the field-of-view (FOV) by registering partially overlapping 3DUS images collected from different probe positions. However, standard registration algorithms frequently encounter difficulties with this task, primarily due to the sector-shaped FOV, which often leads to pronounced local minima, thereby obstructing optimization efforts.To address these limitations, we propose LOTUS, a novel Latent Diffusion Model (LDM) specifically designed for 3DUS FOV outpainting. LOTUS innovatively encodes the 3DUS data into a compact latent space and performs outpainting at test time, effectively extending the sector-shaped FOV into a standard rectangular shape. This transformation facilitates a more robust registration by mitigating the issues of local minima associated with the original FOV shape. Experimental results show that LOTUS significantly improves the accuracy of the registration as well as the efficiency of the outpainting process compared to existing models. The code is available at https://github.com/MedICL-VU/LOTUS.} }
Endnote
%0 Conference Paper %T LOTUS: Latent Outpainting Diffusion Model for Three-Dimensional Ultrasound Stitching %A Xing Yao %A Runxuan Yu %A Nick DiSanto %A Ehsan Khodapanah Aghdam %A Kanyifeechukwu Jane Oguine %A Daiwei Lu %A Ange Lou %A Jiacheng Wang %A Dewei Hu %A Gabriel A Arenas %A Baris Oguz %A Alison Marie Pouch %A Nadav Schwartz %A Brett Byram %A Ipek Oguz %B Proceedings of The 8th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2026 %E Tolga Tasdizen %E Shireen Elhabian %E Ronald Summers %E Chen Chen %E Lisa Koch %E Yan Zhuang %F pmlr-v301-yao26a %I PMLR %P 1740--1754 %U https://proceedings.mlr.press/v301/yao26a.html %V 301 %X 3D ultrasound (3DUS) stitching can enlarge the field-of-view (FOV) by registering partially overlapping 3DUS images collected from different probe positions. However, standard registration algorithms frequently encounter difficulties with this task, primarily due to the sector-shaped FOV, which often leads to pronounced local minima, thereby obstructing optimization efforts.To address these limitations, we propose LOTUS, a novel Latent Diffusion Model (LDM) specifically designed for 3DUS FOV outpainting. LOTUS innovatively encodes the 3DUS data into a compact latent space and performs outpainting at test time, effectively extending the sector-shaped FOV into a standard rectangular shape. This transformation facilitates a more robust registration by mitigating the issues of local minima associated with the original FOV shape. Experimental results show that LOTUS significantly improves the accuracy of the registration as well as the efficiency of the outpainting process compared to existing models. The code is available at https://github.com/MedICL-VU/LOTUS.
APA
Yao, X., Yu, R., DiSanto, N., Aghdam, E.K., Oguine, K.J., Lu, D., Lou, A., Wang, J., Hu, D., Arenas, G.A., Oguz, B., Pouch, A.M., Schwartz, N., Byram, B. & Oguz, I.. (2026). LOTUS: Latent Outpainting Diffusion Model for Three-Dimensional Ultrasound Stitching. Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 301:1740-1754 Available from https://proceedings.mlr.press/v301/yao26a.html.

Related Material