Organ Tracking via Synchronized Motion Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image-guided navigation in minimally invasive medical procedures, such as TAVI, faces challenges due to limited accuracy and operator bias, especially when navigating in moving areas like the thoracic region, and the use of intra-procedural images with reduced quality that may not reveal anatomical structures without contrast liquid, which can be harmful if overused.
Innovation Solution
A method for tracking objects in a target area of a moving organ using synchronized periodic motion patterns between multiple features, where the positions of these features are learned from sequences of image frames, allowing for the derivation of obscured feature positions based on dynamic geometric relations and periodic motion patterns, reducing the need for contrast agents and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If contrast liquid is used to visualize anatomical structures in intra-procedural images, then image quality and visibility of anatomical structures is improved, but patient harm risk increases due to potential overuse and toxicity
Solution Approach 1:
The system creates a digital copy of the anatomical structures by training a deep learning model on pre-procedural images with contrast liquid. The trained model can then generate synthetic contrast-enhanced images from non-contrast intra-procedural images, eliminating the need to administer additional contrast liquid during the procedure while maintaining visualization quality
Solution Approach 2:
The system performs preliminary action by acquiring and processing pre-procedural images with contrast liquid before the actual intervention. This allows the deep learning model to learn the relationship between contrast-enhanced and non-contrast images in advance, so that during the procedure, accurate anatomical visualization can be achieved without administering contrast liquid
2Ease of operation
If manual visual inspection of intra-procedural images is used for navigation, then operator flexibility is maintained, but positioning accuracy decreases due to operator bias and subjective evaluation
Solution Approach 1:
The system replaces the manual visual inspection mechanism with an automated deep learning-based image processing system. The trained neural network automatically identifies anatomical structures and provides quantitative positioning information, eliminating operator bias and subjective evaluation while maintaining ease of use through automated analysis
Solution Approach 2:
The system implements feedback by continuously analyzing intra-procedural images and providing real-time quantitative information about tool position relative to anatomical structures. This automated feedback loop allows operators to make precise adjustments without relying on subjective visual assessment, improving positioning accuracy while maintaining operational flexibility
3Duration of action of stationary object
If tracking is continued when the tracked feature is obscured, then tracking continuity is maintained, but positioning accuracy deteriorates due to accumulation of modeling errors
Solution Approach 1:
When the tracked feature is obscured, the system creates a synthetic copy of the expected feature appearance using the deep learning model and pre-procedural anatomical information. This allows the tracking algorithm to maintain continuity by comparing actual images against the synthetic template, preventing complete tracking failure while limiting error accumulation to only the periods when features are actually obscured
Data Source
AI summary
A method for tracking position of features of a moving organ from at least one sequence of image frames of the moving organ involves identifying at least a first feature and a second feature of the organ in a reference image frame. Positions of the first and second features in other image frames are tracked in order to learn motion patterns of the first and second features. A dynamic geometric relation between the first and second features is determined. In the event that the first feature of the organ is obscured in a given image frame, position of the first feature in the given image frame is determined using position of the second feature in the given image frame and the dynamic geometric relation between the first and second features.