Endoscopic Image Processing for Abnormal Tissue Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques for endoscopic images often incorrectly identify normal tissue edges as abnormal due to shading and shadow effects, particularly near mucosal folds, leading to erroneous determinations.
Innovation Solution
An image processing device and method that detect a deep region within an intraductal image, extract contour edges, analyze convex regions for curvature and direction, and identify abnormal regions by detecting convex directions pointing towards the deep region, using a low-absorption wavelength component and edge surrounding region elimination to suppress false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional edge detection techniques are used to identify abnormal tissue edges, then detection sensitivity is improved, but false positive rate increases due to shading and shadow effects near mucosal folds
Solution Approach 1:
The inner wall contour is divided into multiple convex regions, and each region is independently analyzed for curvature and convex direction. This segmentation allows the system to distinguish between normal folds (where convex directions point away from deep regions) and abnormal lesions (where convex directions point toward deep regions), thereby reducing false positives while maintaining detection sensitivity
Solution Approach 2:
The patent applies different analysis criteria to different parts of the contour based on their local geometric properties. By analyzing the convex direction of each convex region relative to the deep region, the system adapts its detection criteria to local characteristics, improving accuracy without increasing false positives from global shading effects
2Measurement precision
If detailed contour analysis is performed to improve detection accuracy, then measurement precision is improved, but computational complexity increases
Solution Approach 1:
The patent performs preliminary extraction of the inner wall contour and identification of convex regions before detailed analysis. By pre-processing the image to isolate the contour and segment it into convex regions, the system reduces the complexity of subsequent curvature and convex direction analysis, making the detailed examination computationally feasible
Solution Approach 2:
The patent introduces a new dimension of analysis by examining the convex direction of contour regions relative to the deep region. This directional analysis in the radial dimension (from inner wall toward center) simplifies the detection logic compared to analyzing all possible edge characteristics, reducing computational complexity while improving precision
Data Source
AI summary
Example embodiments of the present invention relate to an image processing apparatus. The apparatus may include a processor and memory storing instructions that when executed on the processor cause the processor to perform the operations of detecting a deep region of a duct in an image and extracting a plurality of contour edges of an inner wall of the duct in the image. The apparatus then may identify a plurality of convex regions among the plurality of contour edges, analyze a respective curvature of each of the plurality of convex regions to identify a convex direction for each of the plurality of convex regions, and detect, as an abnormal region, a convex region having a convex direction directed toward the deep region.