Abstract

Cephalometric landmark detection plays a crucial role in Maxillofacial clinical practice, supporting both diagnosis and preoper- ative planning. Recent advances in computer vision have enabled au- tomatic landmark detection in 2D and 3D medical images. However, achieving accurate predictions remains challenging, particularly in 3D medical volumes, due to their high spatial resolution and the substantial variability in voxel spacing, volume size, and scanned anatomical region. In this work, we propose Attention Voting, a cross-attention based mech- anism that models landmark detection as a matching problem between learned landmark embeddings and spatial feature maps, converting at- tention scores into continuous coordinate predictions. We evaluate our method on an in-house dataset of CT and CBCT scans as well as on the public MMLD dataset. Experimental results demonstrate that our method matches or outperforms existing methods in terms of localization accuracy. Code: https://github.com/Chavelanda/attention_voting

Links to Paper and Supplementary Materials

Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/ODIN_006.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=LQjk7eUqUW

BibTex

@InProceedings{MelGia_Attention_MICCAISAT2026,
        author = { Melacini, Giacomo AND Bortolami, Carlotta AND Gulotta, Chiara AND Badiali, Giovanni AND Lisanti, Giuseppe AND Di Stefano, Luigi AND Salti, Samuele},
        title = { { Attention Voting for 3D Cephalometric Landmark Detection } },
        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}
}


back to top