Hemorrhage Edge Detection in Endoscope Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current endoscope image processing methods are inefficient in detecting hemorrhages due to reliance on subjective human interpretation and are burdened by the need to review large numbers of images, leading to prolonged diagnosis times and variable quality.
Innovation Solution
An image processing apparatus that extracts hemorrhage edge candidates based on changes in color signals within small areas of a medical image, calculates feature quantities, and determines the presence of hemorrhages using these changes, thereby reducing the burden on human observers and improving diagnostic accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all captured images are displayed for observation and diagnosis, then comprehensive diagnosis coverage is achieved, but the time required for shadow-reading increases significantly
Solution Approach 1:
The system performs preliminary automated analysis of all captured images before clinical review, pre-identifying potential hemorrhage cases. This preliminary action filters the large volume of images (43200 images over 6 hours) down to only those requiring clinical attention, allowing comprehensive coverage without requiring clinicians to review every single image manually.
Solution Approach 2:
An automated image analysis system acts as an intermediary between image capture and clinical diagnosis. This intermediary process objectively evaluates all images for hemorrhage indicators and presents only relevant findings to clinicians, reducing their workload while maintaining comprehensive diagnostic coverage.
2Reliability
If computer-aided diagnosis is implemented to automatically detect hemorrhages, then diagnostic consistency is improved, but the complexity of the system increases
Solution Approach 1:
The system transforms the complex task of hemorrhage detection into quantitative parameter analysis by measuring color signal changes (RGB values) in image pixels. By converting visual assessment into objective numerical measurements of color intensity and variation, the system achieves consistent diagnostic results while managing complexity through standardized parameter evaluation.
Solution Approach 2:
The system replaces subjective human visual assessment with automated computational analysis. Instead of relying on clinicians' subjective interpretation of image colors and patterns, the system uses algorithmic processing to objectively detect hemorrhage indicators, improving consistency while reducing the complexity burden on human operators.
3Extent of automation
If hue, saturation, and brightness values are compared with preset sample values, then hemorrhage detection is automated, but the determination results depend heavily on the accuracy of preset samples
Solution Approach 1:
Instead of relying on a limited set of preset sample values, the system performs excessive analysis by evaluating multiple color parameters (R, G, B values and their variations) across all pixels in each image. This partial/excessive approach to parameter measurement compensates for the lack of extensive preset samples, maintaining both automation and reliability through comprehensive parameter evaluation.
Solution Approach 2:
The system moves beyond simple hue, saturation, and brightness comparison with preset samples to a more robust parameter-based approach. It measures actual RGB color values and their changes across image pixels, transforming the detection method into one that relies on quantitative parameter variation rather than comparison with potentially inaccurate preset references, thereby maintaining automation while improving determination accuracy.
Data Source
AI summary
A hemorrhage edge candidate area extraction section extracts a candidate area for the outline part of a hemorrhage area, based on an image signal of a medical image constituted by multiple color signals obtained by capturing an image of a living body. A feature quantity calculation section calculates a feature quantity of the hemorrhage area based on calculation of the amount of change in the image signal in a small area including the candidate area, among multiple small areas obtained by dividing the medical image. A hemorrhage edge determination section determines whether or not the candidate areas are the outline part of the hemorrhage area based on the feature quantity.