Abstract
3D shape models have become a popular tool for representing variations within specific domains of 3D shapes, such as faces, in a compact and interpretable space. As such, they can improve our understanding of these domains and reveal how individual morphology varies across specific anatomical regions. Although most existing models target healthy adult populations, several recent works have focused on the distinct anatomical characteristics of infant faces and crania, some also encoding cranial pathologies. However, cleft lip and palate has remained largely neglected. Only one 3D shape model has focused on two types of cleft palate in isolation, without accounting for correlations among different cleft types or their impact on facial morphology. To address this gap, we introduce INCLEFT, the first implicit 3D infant shape model to jointly represent the face, cranium, and palate, capturing healthy anatomical variation alongside most types of cleft lip and palate. The model is built on a dataset comprising incomplete 3D photogrammetric and structured-light scans of both healthy and cleft-affected infants, some of which are complemented by corresponding intraoral scans. We developed a custom annotation scheme that guides a tailored nonrigid surface-registration pipeline based on an artist-designed template mesh that connects the face and palate. Together, the annotation and registration form an essential preprocessing step for training the implicit shape model. To address the challenge of training on incomplete data, we combine several recent methods from research on implicit neural representations and provide a visual evaluation of the model’s capabilities and limitations. We hope that the presented analysis, together with the release of several model configurations, will foster further research on implicit shape models, improve understanding of cleft-related morphology, and ultimately contribute to more standardized treatment approaches.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/ShapeMI_041.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=WBzTjiWVT5
BibTex
@InProceedings{SchTil_An_MICCAISAT2026,
author = { Schnabel, Till N. AND Lill, Yoriko AND Weiherer, Maximilian AND Schenk, Ruben W. G. AND Jeyalingam, Vennilah AND Nava Martinez, Iris AND Benitez, Benito K. AND Keller, Marilyn AND Mueller, Andreas A. AND Solenthaler, Barbara},
title = { { An Implicit 3D Face and Palate Shape Model for Infant Orofacial Clefts } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17259},
month = {pending},
page = {pending}
}