Facial Palsy Simulation Using a Partially Fine-tuned ControlNet with Contour-guided and Organ-size Deformation Conditioning

View at publisher Summarise with VIRAT
Authors: Atsushi Tajima, Masataka Seo, Yen-Wei Chen
Cited by: 0
References: 12

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Facial nerve palsy causes impairment in voluntary facial movements, and its severity in Japan is typically evaluated using the 40-point Yanagihara method. However, visual inspection introduces inter-rater variability, and patient privacy limits the use of real images. In this work, we present a simulation approach that synthesizes facial palsy images using a Stable Diffusion model with a partially fine-tuned ControlNet. Fine-tuning is limited to timesteps corresponding to abstract feature representations, allowing the generation of pathological expressions while suppressing the reproduction of actual identity. Moreover, a preprocessing step modifies the size and position of facial organs and incorporates contour information from non-patient faces, thereby enhancing expression fidelity while ensuring privacy protection. This framework has potential applications in medical training and clinical research.

Fields of Study & Subjects

Generative Adversarial Networks and Image Synthesis Facial Nerve Paralysis Treatment and Research Face recognition and analysis

Publication details

Journal / source
International Conference on Consumer Electronics
Pages
1–5
Publication date
3 February 2026
Language
EN