MICCAI STSR 2025 Challenge: Semi-Supervised Teeth and Pulp Segmentation and CBCT-IOS Registration

2025-12-02

AI Summary

In digital dentistry, the scarcity of annotated data for cone-beam computed tomography (CBCT) and intraoral scanning (IOS) impedes automated root canal segmentation and cross-modal registration. Method: We propose the first systematic semi-supervised learning benchmark for tooth-pulp segmentation and CBCT-IOS registration, releasing a large-scale dataset with both labeled and unlabeled samples. Our approach integrates nnU-Net with Mamba-inspired state-space models, augmented by pseudo-labeling, consistency regularization, a differentiable PointNetLK–SVD registration framework, and geometric augmentation. Results: The method achieves a Dice score of 0.967 for pulp segmentation and an instance-level association F1 score of 0.738; cross-modal registration attains high accuracy even with minimal labeled data. All top-performing methods are open-sourced and fully reproducible, establishing a standardized evaluation protocol and facilitating clinical deployment of semi-supervised learning in digital dentistry.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Semi-Supervised Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Cone-Beam Computed Tomography (CBCT) and Intraoral Scanning (IOS) are essential for digital dentistry, but annotated data scarcity limits automated solutions for pulp canal segmentation and cross-modal registration. To benchmark semi-supervised learning (SSL) in this domain, we organized the STSR 2025 Challenge at MICCAI 2025, featuring two tasks: (1) semi-supervised segmentation of teeth and pulp canals in CBCT, and (2) semi-supervised rigid registration of CBCT and IOS. We provided 60 labeled and 640 unlabeled IOS samples, plus 30 labeled and 250 unlabeled CBCT scans with varying resolutions and fields of view. The challenge attracted strong community participation, with top teams submitting open-source deep learning-based SSL solutions. For segmentation, leading methods used nnU-Net and Mamba-like State Space Models with pseudo-labeling and consistency regularization, achieving a Dice score of 0.967 and Instance Affinity of 0.738 on the hidden test set. For registration, effective approaches combined PointNetLK with differentiable SVD and geometric augmentation to handle modality gaps; hybrid neural-classical refinement enabled accurate alignment despite limited labels. All data and code are publicly available at https://github.com/ricoleehduu/STS-Challenge-2025 to ensure reproducibility.

Problem

• Addresses scarcity of annotated dental CBCT and IOS data for automated analysis
• Develops semi-supervised methods for teeth and pulp canal segmentation in CBCT scans
• Creates cross-modal registration techniques aligning CBCT and IOS data with limited labels

Innovation

• Semi-supervised learning with pseudo-labeling and consistency regularization
• Combining PointNetLK with differentiable SVD for cross-modal registration
• Hybrid neural-classical refinement to handle limited labeled data