Endoscopic Navigation via Image Registration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current endoscopic navigation techniques face challenges such as incorrect path selection and difficulty in identifying airways in CT slices, leading to navigation errors and variations in skill level among physicians, especially in complex airway trees like the tracheobronchial tree, which can result in missed diagnoses and invasive follow-up procedures.
Innovation Solution
A system for continuous real-time image-based guidance during endoscopy, utilizing pre-computed 3D image data to register and track the endoscope's position relative to predefined routes and regions of interest, superimposing 3D routes and ROIs onto live endoscopic images, allowing for continuous guidance and reducing reliance on mental reconstruction and external sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If physicians mentally reconstruct 3D anatomy from 2D CT slices to plan routes, then route planning can be performed without additional hardware, but navigation errors increase and path selection accuracy deteriorates
Solution Approach 1:
The patent creates a virtual bronchoscopic copy of the patient's airway tree from pre-operative CT images. This virtual model is then registered with real-time endoscopic video to provide visual guidance along the planned path. The copy allows physicians to see the intended route and anatomical landmarks without mentally reconstructing 3D structures from 2D slices, thereby improving path selection accuracy while maintaining procedural simplicity
Solution Approach 2:
The patent introduces a computer-based navigation system as an intermediary between the CT images and the endoscopic procedure. This system automatically generates 3D reconstructions, plans optimal routes, and provides real-time visual feedback by overlaying virtual anatomical structures on endoscopic video. The intermediary eliminates the need for manual mental reconstruction while significantly improving navigation precision
2Measurement precision
If E/M guidance techniques are used to locate the bronchoscope, then navigation precision improves, but device complexity increases and procedural flexibility decreases
Solution Approach 1:
The patent replaces the mechanical E/M tracking system with a computer vision-based image registration system. Instead of using electromagnetic sensors in the bronchoscope, the system uses image processing to track the bronchoscope's position by comparing real-time endoscopic video with pre-acquired 3D CT images. This substitution eliminates complex hardware while maintaining or improving location accuracy
Solution Approach 2:
The patent makes the navigation system adaptable to different bronchoscope types and procedural needs without requiring specialized hardware. The image-based approach works with standard endoscopic equipment, allowing the same navigation system to be used across various procedures and device configurations, thereby reducing overall system complexity and increasing procedural flexibility
3Loss of information
If the E/M probe occupies the instrument channel, then navigation information is provided, but the ability to perform critical procedural steps deteriorates
Solution Approach 1:
The patent extracts the navigation function from the instrument channel by using the endoscopic video feed itself for tracking. The E/M probe is removed from the bronchoscope, and navigation information is obtained through image processing of the video signal. This extraction allows the instrument channel to remain free for performing critical procedural steps like biopsies while continuously providing location information through the navigation system
Solution Approach 2:
The patent enables the endoscopic video system to serve dual purposes: both providing visual guidance for navigation and enabling navigation tracking. The same video feed that shows the anatomical landscape is also used by the image registration algorithm to track bronchoscope position. This self-service approach eliminates the need for separate sensing hardware, maintaining procedural flexibility while continuously providing location information
4Ease of operation
If image-based guidance with virtual bronchoscopic navigation is used, then ease of operation improves, but real-time feedback capability deteriorates
Solution Approach 1:
The patent performs comprehensive 3D reconstruction, airway segmentation, and route planning during the pre-operative phase based on CT images. By completing these computationally intensive tasks beforehand, the system minimizes real-time processing requirements. During the procedure, only lightweight image registration and video overlay operations are needed, enabling real-time feedback while maintaining high-quality navigation guidance
Solution Approach 2:
The patent segments the navigation process into distinct phases: pre-operative 3D reconstruction and route planning, and intra-procedural real-time image registration. This segmentation allows computationally intensive tasks to be performed offline with sufficient computing resources, while the real-time component uses optimized algorithms that run at video frame rates, thereby achieving both high guidance quality and real-time performance
Data Source
AI summary
Methods and apparatus provide continuous guidance of endoscopy during a live procedure. A data-set based on 3D image data is pre-computed including reference information representative of a predefined route through a body organ to a final destination. A plurality of live real endoscopic (RE) images are displayed as an operator maneuvers an endoscope within the body organ. A registration and tracking algorithm registers the data-set to one or more of the RE images and continuously maintains the registration as the endoscope is locally maneuvered. Additional information related to the final destination is then presented enabling the endoscope operator to decide on a final maneuver for the procedure. The reference information may include 3D organ surfaces, 3D routes through an organ system, or 3D regions of interest (ROIs), as well as a virtual endoscopic (VE) image generated from the precomputed data-set. The preferred method includes the step of superimposing one or both of the 3D routes and ROIs on one or both of the RE and VE images. The 3D organ surfaces and routes may correspond to the surfaces and paths of a tracheobronchial airway tree extracted, for example, from 3D MDCT images of the chest.