Virtual Colonoscopy Topological Support Tree
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional colonoscopy procedures are invasive, costly, and uncomfortable, and non-invasive virtual colonoscopies face challenges in accurately identifying the inner wall of the colon, especially at air-tagged region interfaces, leading to potential leakage and misevaluation of colonic lesions.
Innovation Solution
A method for determining the topological support of a tubular structure without relying on segmentation parameters, using region growing techniques to build a tree representation of the structure, which allows for accurate 3D representation and identification of the colon's inner wall without requiring interface detection, suitable for various image data types and scanning devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If conventional segmentation methods are used to identify the colon's inner wall, then the procedure can be automated, but the accuracy deteriorates at air-tagged region interfaces leading to leakage
Solution Approach 1:
The patent introduces an intermediary substance (e.g., barium-based contrast material or oral contrast) that fills the colon lumen to create a distinct radiodensity difference between the colon interior and exterior. This intermediary enables automatic segmentation algorithms to reliably identify the colon's inner wall by detecting the interface between regions of different densities, thereby resolving the accuracy problem at air-tagged region interfaces while maintaining automation.
2Ease of manufacture
If interface detection methods are used to locate the colon's inner wall, then segmentation can be performed, but reliability deteriorates when interfaces are thick or inhomogeneous
Solution Approach 1:
The patent changes the physical parameter of the colon contents by introducing contrast material that alters the radiodensity characteristics. This transforms the interface from a thick or inhomogeneous air-tagged region boundary into a sharp, well-defined interface between contrast-filled lumen and surrounding tissues. The modified density parameter enables reliable automatic detection even in collapsed colons or cases with obstructive tumors, significantly improving segmentation reliability.
3Object-affected harmful factors
If traditional virtual colonoscopy methods are used, then non-invasive examination is achieved, but processing time increases due to complex segmentation requirements
Solution Approach 1:
The patent performs preliminary action by having the patient ingest contrast material before the CT scan, which pre-prepares the colon lumen with a substance that creates optimal radiodensity contrast. This preliminary preparation eliminates the need for complex post-processing segmentation algorithms to deal with ambiguous air-tagged region interfaces, thereby reducing computational processing time while maintaining the non-invasive nature of the procedure.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for determining an estimation of a topological support of a tubular based structure comprising an inner wall and a plurality of distinct regions, the method comprising (a) obtaining image data representative of the tubular based structure; (b) placing an initial seed in an initial region selected from one of the distinct regions; (c) performing an initial region growing until an initial resulting area comprises at least a portion of the inner wall and at least a portion of a neighboring region corresponding to one of the distinct regions; (d) starting a tree comprising an initial tree node corresponding to the initial region; (e) for each neighboring region: placing a subsequent seed in the neighboring region; performing a corresponding subsequent region growing until a subsequent resulting area comprises at least a portion of the inner wall and at least a portion of an additional neighboring region; and adding a tree node corresponding to the neighboring region in the tree; (f) performing processing step (e) for each of the additional neighboring regions; and (g) filtering the tree according to predetermined topological parameters to thereby determine the estimation of the topological support of the tubular based structure. Applications of the method for estimating a colon topology for virtual colonoscopy are also disclosed.