Endoscopic Diagnosis Support Using L*a*b* Tone Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current endoscopic diagnosis methods face challenges in efficiently and accurately identifying bleeding regions from a large number of images captured by endoscopes, particularly due to the reliance on preset sample values which can lead to inconsistent discernment results.
Innovation Solution
An endoscopic diagnosis support method that calculates tone from color signals of divided image zones and discerns bleeding regions by judging differences among these zones, allowing for precise identification of bleeding areas within endoscopic images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If preset sample values of hue, saturation and brightness are used to identify bleeding regions, then the diagnosis process can be automated, but the discernment results become inconsistent and unreliable
Solution Approach 1:
The patent transforms the color discrimination approach from using fixed preset values to using dynamic parameter relationships. Specifically, it converts hue, saturation, and brightness values into a different parameter space (L*a*b* color space) and uses ratio relationships between these parameters rather than absolute threshold values. This allows the system to maintain automation while improving reliability by adapting to variations in imaging conditions through normalized parameter comparisons.
Solution Approach 2:
The patent introduces an intermediary transformation process that converts the raw color signals into a standardized color space (L*a*b*) before performing discrimination. This intermediary step acts as a mediator that decouples the discrimination logic from the specific imaging conditions, allowing consistent results across different scenarios while maintaining automated operation.
2Area of stationary object
If a large number of images are captured during the observation period, then comprehensive coverage of the observation region is achieved, but the time required for diagnosis increases significantly
Solution Approach 1:
The patent transforms the discrimination criteria from multiple color parameters (hue, saturation, brightness) into a simplified parameter relationship based on ratios in the L*a*b* color space. This parameter transformation enables faster processing by reducing the computational complexity of comparing multiple threshold values across numerous images, thereby decreasing diagnosis time while maintaining comprehensive coverage.
Solution Approach 2:
The patent divides the image analysis into discrete processable units by evaluating each image independently using the parameter ratio method. This segmentation allows for efficient automated processing of large numbers of images without requiring manual review of each one, significantly reducing diagnosis time while maintaining comprehensive observation region coverage.
3Ease of manufacture
If conventional color discrimination methods are used, then the implementation is simple, but the accuracy of bleeding region identification is insufficient
Solution Approach 1:
The patent maintains simplicity by using a standardized color space transformation (L*a*b*) as an intermediary step that is computationally straightforward. This intermediary approach preserves ease of implementation while dramatically improving accuracy by transforming the color data into a space where bleeding regions can be more reliably distinguished from normal tissue using ratio-based criteria.
Solution Approach 2:
The patent improves accuracy by changing the parameter representation from raw RGB or HSV values to L*a*b* color space parameters, and further to ratio relationships between these parameters. This parameter transformation enhances the separability of bleeding regions from normal tissue, improving identification accuracy while maintaining computational simplicity through standardized transformations.
Data Source
AI summary
Provided are an endoscopic diagnosis support method, an endoscopic diagnosis support apparatus, and an endoscopic diagnosis support program, all of which are capable of extracting an image picking up a bleeding region easily and accurately from among a large number of endoscopic images picked up by an endoscope observation apparatus by calculating a tone from a color signal of each of plural image zones obtained by dividing the endoscopic image; and discerning an image zone including a bleeding region by judging a difference among each of the plural image zones based on a tone of the calculated each image zone in the endoscopic diagnosis support apparatus for supporting an endoscopic diagnosis performed based on an endoscopic image picked up by an endoscope observation apparatus.