Abstract
Understanding normative paediatric facial development, both globally and regionally, provides clinicians with a model of normal growth to compare with craniofacial syndromes’ facial development. This helps assess severity, plan surgical approaches, guide clinical decisions and support family communication. Whilst models for 3D face anatomies are considerably advanced, disentangling the factors that shape face changes remains challenging, with existing methods that separate the identity of different facial regions overlooking age, a critical factor in paediatric populations where facial morphology changes rapidly with growth. Thus, we propose a Swapped and Age-Disentangled Variational Autoencoder for healthy 0-17 years 3D faces that disentangles age from identity. Building on region-wise latent disentanglement through feature-swapping, we introduce an age factor and isolate it from identity, globally and within each facial region, without requiring longitudinal data. This provides a normative, per-region reference of age-associated shape variations in paediatric faces and a foundation for extension to syndromic populations.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/PedAItrics_012.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
BibTex
@InProceedings{ReiAth_RegionWise_MICCAISAT2026,
author = { Reissis, Athena AND Smith, Luke AND Schievano, Silvia AND Foti, Simone},
title = { { Region-Wise Age Progression in Disentangled VAEs for 3D Face Meshes } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17257},
month = {pending},
page = {pending}
}