Intra-luminal Image Processing for Lesion Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies face challenges in accurately detecting specific regions, such as shadows, halations, grooves, residues, and bubbles, within intra-luminal images, as they often include non-target features like shadows and grooves, which can lead to incorrect identification of abnormal regions.
Innovation Solution
An image processing device and method that includes a non-target region detecting unit, a pixel-of-interest region setting unit, a surrounding region determining unit, a reference plane forming unit, and an outlier pixel detecting unit, which identifies and excludes non-target regions like shadows and grooves to form a reference plane based on surrounding information, allowing for accurate detection of outlier pixels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pixel value change amounts and circumferential pixel value change amounts are calculated to detect lesion candidate regions, then lesion detection capability is improved, but false positive detection increases due to non-target regions like shadows and grooves
Solution Approach 1:
The patent extracts and removes non-target regions (shadows, halations, grooves, residues, bubbles) from the image data before performing pixel value analysis. By separating these interfering elements from the actual tissue structures, the system can accurately detect lesion candidate regions without false positives caused by non-target features.
Solution Approach 2:
The patent segments the image into target regions (tissue structures) and non-target regions (artifacts). This segmentation allows the system to process only the relevant tissue areas for lesion detection while excluding artifacts that would cause false positive results.
2Measurement precision
If a reference plane is formed using surrounding region information to detect outlier pixels, then detection accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent applies local quality analysis by examining pixel values and their surroundings at specific locations. The reference plane is formed using only the immediate surrounding region of each pixel, allowing the system to detect local outliers while maintaining manageable processing complexity through localized rather than global analysis.
Data Source
AI summary
An image processing device includes: a non-target region detecting unit that detects a region that is not to be examined as a non-target region from an image; a pixel-of-interest region setting unit that sets a pixel-of-interest region in a predetermined area including a pixel-of-interest position in the image; a surrounding region determining unit that determines a surrounding region, which is an area for acquiring information for use in forming a reference plane with respect to the pixel-of-interest position, based on the non-target region; a reference plane forming unit that forms the reference plane based on the information in the surrounding region; and an outlier pixel detecting unit that detects an outlier pixel having a pixel value numerically distant from circumferential values based on a difference between corresponding quantities of the reference plane at each pixel position and of the original image.