Abstract
Accurate registration of Intraoral Scans (IOS) with Cone- Beam Computed Tomography (CBCT) enables integration of dental crown surfaces with the tooth roots and the surrounding alveolar bone. However, IOS-CBCT registration remains challenging because of lim- ited anatomical overlap, cross-modal differences, and unreliable corre- spondences. Task 2 of the MICCAI STS 2026 Challenge formulates this setting as a semi-supervised rigid registration problem, in which IOS meshes must be aligned with their corresponding CBCT volumes. For this task, we propose a landmark-driven coarse-to-fine framework in which modality-specific networks predict corresponding dental landmarks from the IOS mesh and CBCT volume. A confidence-weighted RANSAC- Kabsch procedure estimates a robust initial transformation, later refined using point-to-plane Iterative Closest Point. On the challenge valida- tion set, the method achieved a Mean Translation Error of 4.988 mm and a Mean Rotation Error of 1.421◦, ranking first on the registration leaderboard at the time of evaluation. The code is available on GitHub: https://github.com/AImageLab-zip/L2L-Registration
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/ODIN_challenges_004.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=y6t967Bmxj
BibTex
@InProceedings{VerAle_LandmarkGuided_MICCAISAT2026,
author = { Veronese, Alex AND Lugli, Matteo AND Carpentiero, Omar AND Marchesini, Kevin AND Lumetti, Luca AND Bolelli, Federico},
title = { { Landmark-Guided Coarse-to-Fine Registration of Intraoral Scans and Cone-Beam CT } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17270},
month = {pending},
page = {pending}
}