Endoscopic Device Positioning in Branched Anatomy via Image Registration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image-guided surgery systems face challenges in accurately tracking and registering medical devices within branched anatomical structures due to movement and tracking inaccuracies, leading to registration errors and misalignment, especially in structures like the lungs or heart.
Innovation Solution
A medical system and method that utilizes a computer model of the anatomical structure and a processor to determine the position of a medical device by comparing image information from its distal end with the model, employing blob detection and tracking algorithms to maintain registration and correct for errors in real-time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If electromagnetic tracking devices and dynamic referencing techniques are used to track medical devices in branched anatomical structures, then tracking capability is provided, but registration accuracy deteriorates due to anatomical motion and tracking inaccuracies
Solution Approach 1:
The system continuously compares captured images from the medical device with corresponding images from the computer model, uses blob detection and tracking to monitor feature points, calculates registration errors in real-time, and provides feedback to correct misalignment between the medical device position and the computer model, thereby maintaining accurate registration despite anatomical motion
Solution Approach 2:
The system creates a virtual copy of the anatomical structure as a computer model with corresponding images and blob features, then compares this copy with real-time captured images from the medical device to determine position and correct registration errors, replacing reliance on inaccurate electromagnetic tracking with image-based virtual model comparison
2Measurement precision
If complex image-based tracking methods with sequential Monte Carlo sampling are used to improve accuracy, then measurement precision improves, but computational complexity increases making real-time application impossible
Solution Approach 1:
The system extracts only the essential blob features from images and tracks their positions and movements through the sequence of captured images, rather than performing full Sequential Monte Carlo sampling of all image data. This extraction of key features maintains tracking accuracy while dramatically reducing computational complexity to enable real-time processing
Solution Approach 2:
Instead of performing complete and computationally intensive Sequential Monte Carlo sampling, the system applies partial action by using simplified blob detection and tracking algorithms that process only the most critical feature points, achieving sufficient accuracy for real-time applications without the full computational burden
3Measurement precision
If traditional registration methods are used prior to medical procedure, then initial registration is achieved, but registration errors develop during procedure due to anatomical structure movement
Solution Approach 1:
The system performs continuous image capture and comparison throughout the medical procedure, continuously updating the registration between the medical device position and the computer model. This continuous action maintains registration stability despite anatomical motion, replacing the traditional single pre-procedure registration step with ongoing real-time registration maintenance
Solution Approach 2:
The system performs preliminary blob detection and feature extraction from both captured images and computer model images before comparing them for position determination. This preliminary processing of key features enables rapid continuous comparison and registration correction throughout the procedure, maintaining stability without requiring full image processing each time
Data Source
AI summary
Information extracted from sequential images captured from the perspective of a distal end of a medical device moving through an anatomical structure are compared with corresponding information extracted from a computer model of the anatomical structure. A most likely match between the information extracted from the sequential images and the corresponding information extracted from the computer model is then determined using probabilities associated with a set of potential matches so as to register the computer model of the anatomical structure to the medical device and thereby determine the lumen of the anatomical structure which the medical device is currently in. Sensor information may be used to limit the set of potential matches. Feature attributes associated with the sequence of images and the set of potential matches may be quantitatively compared as part of the determination of the most likely match.