Related Experiment Video
Updated: Jun 21, 2026
07:13
Multimodal Cross-Device and Marker-Free Co-Registration of Preclinical Imaging Modalities
Published on: October 27, 2023
Selective image similarity measure for bronchoscope tracking based on image registration
Daisuke Deguchi1, Kensaku Mori, Marco Feuerstein
1Graduate School of Information Science, Nagoya University, Furo-cho, Chikusa-ku, Nagoya-shi, Aichi 464-8603, Japan. ddeguchi@is.nagoya-u.ac.jp
Medical Image Analysis
|July 14, 2009
Summary
This study introduces a new image similarity method for bronchoscope navigation. It accurately tracks the bronchoscope without external sensors by analyzing characteristic image structures, improving upon traditional methods.
Area of Science:
- Medical Imaging
- Computer Vision
- Robotics
Background:
- Camera motion tracking is crucial for image-guided therapies.
- Existing tracking methods often rely on electromagnetic sensors, which are difficult to integrate into ultra-thin bronchoscopes.
- Global image similarity measures fail in scenes with limited characteristic structures.
Purpose of the Study:
- To develop a selective image similarity measurement method for enhanced bronchoscope navigation.
- To enable sensorless camera tracking for flexible endoscopes, particularly bronchoscopes.
- To improve the robustness of image registration in challenging visual environments.
Main Methods:
- A novel method for computing image similarities based on characteristic structure extraction.
- Image analysis involves dividing images into subblocks and selecting those with salient features.
- Image similarity is calculated exclusively within these selected subblocks.
Main Results:
- The proposed method successfully tracked up to 1600 consecutive bronchoscopic images (approx. 50s) without external position sensors.
- Tracking performance showed significant improvement compared to standard methods using global image similarity.
- Demonstrated application in flexible endoscope navigation, specifically for bronchoscope navigation systems.
Conclusions:
- The selective image similarity method offers a robust solution for sensorless bronchoscope tracking.
- This approach overcomes limitations of traditional methods in environments with sparse features.
- Enables more effective image-guided interventions using flexible endoscopes.
