Abstract
Transferring pretrained registration models to ultrasound is challenging because of weak cross-modal correspondence, limited field of view, and large initial pose differences. We propose US-multiGradICON, a global-to-local framework for CT–ultrasound and MR–ultrasound registration. Surface-based PCA–ICP initialization provides robust global alignment, after which pretrained multiGradICON is finetuned with image-similarity, gradient inverse-consistency, and Dice supervision to predict the final deformation field in a single feed-forward pass. We evaluate the method on kidney CT–ultrasound and prostate MRI–ultrasound registration. US-multiGradICON outperformed all competing methods across both applications. Compared with the strongest baseline, it significantly reduces target registration error from to for kidney registration and from to for prostate registration. These results demonstrate that robust surface initialization and anatomical supervision enable effective transfer of pretrained multimodal registration models to distinct CT/MR–ultrasound applications.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/ASMUS_058.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=2kVEXDMhLW
BibTex
@InProceedings{d’AGab_Global_MICCAISAT2026,
author = { d’Albenzio, Gabriella AND Liu, Mengting AND Ma, Shixing AND Yang, Chunna AND Wei, Yuhao AND Vasu, Shrisharanyan AND Vijesh, Aniketh AND Ma, Xihan AND Taylor, Zachary P. AND Pias, Tanmoy Sarkar AND Demir, Başar AND Gyöngy, Miklós AND Ceran, Yasin AND Niethammer, Marc AND Kapur, Tina AND Aylward, Stephen AND Fichtinger, Gabor AND Min, Zhe AND Rusu, Mirabela},
title = { { Global to Local Registration: Transferring Pretrained Registration Models to CT/MR–Ultrasound via Anatomical Supervision } },
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}
}