Geometry-Aware Bronchoscopy Navigation
I began with developing bronchoscopy localization methods to connect endoscopic observations with pre-operative airway structure.
About Me
I am an applied scientist in medical AI at Ant Group, where I work with Dr. Le Lu. Before that, I received my Ph.D. from the Institute of Automation, Chinese Academy of Sciences (CASIA) in June 2026, advised by Professor Hongbin Liu. My research focuses on AI for healthcare, medical imaging, and foundation models, with recent work on bronchoscopy localization, endoscopic depth estimation, and robot-assisted surgery.
I am interested in building reliable learning systems that connect medical visual perception, clinical navigation, and deployable intelligent healthcare tools.
Research Interests
Research Arc
Generalizable Endoscopic Perception
I moved toward foundation models that transfer across patients, datasets, and clinical environments for endoscopy-related clinical support.
Vision-language Models for Healthcare
My recent research explores vision-language systems that can support more reliable interventional healthcare workflows.
Research Project: Vision Navigation for Interventional Surgeries
CASIA
Developed image and video analysis methods for endoscopy-related clinical support, including airway scene understanding, depth estimation, and bronchoscopy localization to help identify the position of the bronchoscope within the airway and support airway examination during clinical procedures.
Education
Institute of Automation, Chinese Academy of Sciences
Ph.D. in Control Theory and Control Engineering
North China Electric Power University
B.E. in Automation
Publications
Selected Publications
2026
BREATH-VL: Vision-Language-Guided 6-DoF Bronchoscopy Localization via Semantic-Geometric Fusion
arXiv preprint arXiv:2601.03713
Snapshot: Builds a hybrid localization pipeline that turns airway anatomy and motion history into language prompts for landmark detection, branch localization, and insertion-depth reasoning, then combines those semantic cues with feature registration for 6-DoF pose estimation.
2025
EndoMamba: An Efficient Foundation Model for Endoscopic Videos via Hierarchical Pre-training
International Conference on Medical Image Computing and Computer-Assisted Intervention Oral
Snapshot: Designs a real-time endoscopic video backbone with bidirectional Mamba blocks for frame-level spatial modeling and temporal Mamba reasoning for online streams, then pretrains it with masked reconstruction and feature alignment for reusable surgical video representations.
2024
EndoOmni: Zero-shot Cross-dataset Depth Estimation in Endoscopy by Robust Self-learning from Noisy Labels
arXiv preprint arXiv:2409.05442
Snapshot: Trains a depth foundation model for endoscopy with a teacher-student self-learning pipeline that estimates confidence in noisy labels and down-weights unreliable pixels through a weighted scale-and-shift invariant loss.
2024
BronchoTrack: Airway Lumen Tracking for Branch-Level Bronchoscopic Localization
IEEE Transactions on Medical Imaging 44 (3), 1321-1333
Snapshot: Introduces multi-object tracking of airway lumens to maintain temporal identity through fast camera motion, then associates tracked detections with a semantic airway graph for branch-level localization without patient-specific retraining.
2024
PANS: Probabilistic Airway Navigation System for Real-Time Robust Bronchoscope Localization
International Conference on Medical Image Computing and Computer-Assisted Intervention
Snapshot: Frames bronchoscope localization as probabilistic pose estimation, propagating pose hypotheses with depth-based motion inference and scoring them with bronchial semantic analysis to improve robustness in visually ambiguous branches.
All Publications
Geometry-Aware Camera Localization for Bronchoscopy
ACM International Conference on Multimedia (ACM MM 2026)
BREATH-VL: Vision-Language-Guided 6-DoF Bronchoscopy Localization via Semantic-Geometric Fusion
arXiv preprint arXiv:2601.03713
Harnessing Foundation Models for Robust and Generalizable 6-DOF Bronchoscopy Localization
MICCAI CREATE 2025, 127
EndoMatcher: Generalizable Endoscopic Image Matcher via Multi-Domain Pre-training for Robot-Assisted Surgery
arXiv preprint arXiv:2508.05205
BronchoTrack: Airway Lumen Tracking for Branch-Level Bronchoscopic Localization
IEEE Transactions on Medical Imaging 44 (3), 1321-1333
DD-VNB: A Depth-Based Dual-Loop Framework for Real-Time Visually Navigated Bronchoscopy
2024 IEEE/RSJ International Conference on Intelligent Robots and Systems Oral
PANS: Probabilistic Airway Navigation System for Real-Time Robust Bronchoscope Localization
International Conference on Medical Image Computing and Computer-Assisted Intervention
BronchoCopilot: Towards Autonomous Robotic Bronchoscopy via Multimodal Reinforcement Learning
2024 IEEE/RSJ International Conference on Intelligent Robots and Systems
Multi-Stage Airway Segmentation in Lung CT Based on Multi-scale Nested Residual UNet
CoRR
Enhancing Bronchoscopy Depth Estimation through Synthetic-to-Real Domain Adaptation
ICRA 2024 Workshop on C4SR+
Services
MICCAI 2026, IROS 2026
IEEE Transactions on Medical Imaging (TMI), IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
News Updates
- Aug 2026 Co-authored paper “EndoMatcher: Generalizable Endoscopic Image Matcher via Multi-Domain Pre-training for Robot-Assisted Surgery” accepted by IEEE Transactions on Medical Imaging (TMI).
- Jul 2026 Co-authored paper “Geometry-Aware Camera Localization for Bronchoscopy” accepted by ACM MM 2026.
- Jul 2026 Joined Ant Group Healthcare.
- Jun 2026 Completed Ph.D. study at the Institute of Automation, Chinese Academy of Sciences.
Profiles & Links
- Scholar Google Scholar profile
- Email tqy00288@gmail.com
- CV Qingyao Tian CV