Bronchoscopy Camera Pose Estimation Using CT Depth Map Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing EM-based localization systems for medical instruments in luminal networks suffer from inaccuracies due to ferromagnetic interference and respiratory motion, leading to unreliable instrument pose estimation, which can cause depth mismatch and navigation failures.
Innovation Solution
A vision-based approach is employed to estimate 6DoF camera poses using video image data and pre-operative CT scans, generating a first depth map and multiple second depth maps to identify the most similar model, thereby improving instrument localization within the luminal network.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If EM-based localization is used for instrument tracking, then real-time 6DoF pose estimation is achieved, but accuracy deteriorates due to ferromagnetic interference and respiratory motion
Solution Approach 1:
The system segments the localization task into multiple independent components: EM field-based tracking for real-time position data, vision-based depth map generation for spatial context, and model-based matching for pose estimation. Each component operates independently to compensate for the weaknesses of others, with EM tracking providing temporal continuity and vision providing spatial accuracy.
Solution Approach 2:
The system merges multiple localization modalities (EM field tracking, vision-based depth mapping, and pre-operative CT model matching) into a unified navigation framework. The EM tracker provides real-time position data while the vision system generates depth maps that are matched against pre-operative models, combining the strengths of different sensing approaches to overcome individual limitations.
2Speed
If EM field tracking is used for navigation, then real-time position data is obtained, but measurement accuracy deteriorates due to ferromagnetic interference from metal instruments and CT scanners
Solution Approach 1:
The system introduces vision-based depth map generation and model matching as intermediary steps between EM field tracking and final pose estimation. The EM tracker provides initial position estimates, which are then refined by matching depth maps against pre-operative CT models, using the model as an intermediary reference frame to correct EM measurement errors caused by ferromagnetic interference.
3Duration of action of moving object
If EM tracking is used for instrument localization, then continuous position monitoring is achieved, but accuracy deteriorates due to respiratory motion effects
Solution Approach 1:
The system performs preliminary action by generating pre-operative 3D models from CT scans before the actual procedure. These pre-acquired anatomical models serve as reference frameworks that remain static and accurate, allowing the system to compensate for respiratory motion by matching real-time depth maps against the pre-established anatomical baseline, thereby maintaining measurement precision throughout continuous tracking.
Data Source
AI summary
Methods and systems provide improved navigation through tubular networks such as lung airways by providing improved estimation of location and orientation information of a medical instrument (e.g., an endoscope) within the tubular network. Various input data such as image data and CT data, are used to model the tubular networks, and the model information is used to generate a camera pose representing a specific site location within the tubular network and/or to determine navigation information including position and orientation for the medical instrument.