UniDCF: A Foundation Model for Comprehensive Dentocraniofacial Hard Tissue Reconstruction

2025-08-15

AI Summary

Craniofacial hard-tissue reconstruction suffers from poor generalizability, low anatomical fidelity, and weak cross-modal adaptability. To address these challenges, we propose the first unified foundation model framework specifically designed for craniofacial hard tissues, integrating point clouds and multi-view images to enable multi-tissue, multi-modal collaborative reconstruction. Our method introduces three key innovations: (1) a multi-modal fusion encoder, (2) score-based denoising optimization, and (3) large-scale joint modeling of point clouds and images—jointly preserving surface smoothness and fine geometric details. Evaluated on 6,609 clinical cases, our approach achieves significantly higher geometric accuracy and structural integrity than state-of-the-art methods, reduces reconstruction time by 99%, and attains a clinical acceptability rate of 94.3%.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Dentocraniofacial hard tissue defects profoundly affect patients' physiological functions, facial aesthetics, and psychological well-being, posing significant challenges for precise reconstruction. Current deep learning models are limited to single-tissue scenarios and modality-specific imaging inputs, resulting in poor generalizability and trade-offs between anatomical fidelity, computational efficiency, and cross-tissue adaptability. Here we introduce UniDCF, a unified framework capable of reconstructing multiple dentocraniofacial hard tissues through multimodal fusion encoding of point clouds and multi-view images. By leveraging the complementary strengths of each modality and incorporating a score-based denoising module to refine surface smoothness, UniDCF overcomes the limitations of prior single-modality approaches. We curated the largest multimodal dataset, comprising intraoral scans, CBCT, and CT from 6,609 patients, resulting in 54,555 annotated instances. Evaluations demonstrate that UniDCF outperforms existing state-of-the-art methods in terms of geometric precision, structural completeness, and spatial accuracy. Clinical simulations indicate UniDCF reduces reconstruction design time by 99% and achieves clinician-rated acceptability exceeding 94%. Overall, UniDCF enables rapid, automated, and high-fidelity reconstruction, supporting personalized and precise restorative treatments, streamlining clinical workflows, and enhancing patient outcomes.

Problem

• Reconstructing multiple dentocraniofacial hard tissues from multimodal inputs
• Overcoming limitations of single-tissue models with poor generalizability
• Addressing trade-offs between anatomical fidelity and computational efficiency

Innovation

• Unified multimodal fusion encoding for reconstruction
• Score-based denoising module enhances surface smoothness
• Largest multimodal dataset enables comprehensive training