Fundus Image Analysis Using Disc Rim Features for Glaucoma Screening
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for diagnosing glaucoma through fundus image analysis are subjective and lack objectivity, leading to inaccurate results and high time consumption.
Innovation Solution
A method involving machine learning models to extract and analyze optic disc, optic cup, and disc rim images from fundus images, using a series of machine learning models to determine glaucoma classification based on integrated features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ophthalmologist uses naked eye or photographing device to estimate cup-disc ratio, then diagnosis can be made, but the estimation is highly subjective and lacks objectivity based on data
Solution Approach 1:
The patent replaces the mechanical/visual estimation method (naked eye or simple photographing device) with an automated image processing system that uses machine learning models. The system extracts optic disc and optic cup images, calculates cup-disc ratio automatically, and provides objective data-based diagnosis, eliminating subjective human estimation while improving measurement precision.
Solution Approach 2:
The system enables self-service diagnosis by automatically processing fundus images through multiple machine learning models to extract relevant features and determine glaucoma status without requiring ophthalmologist's manual measurement. The automated system performs the diagnostic function independently, providing consistent and objective results.
2Productivity
If ophthalmologist manually observes and judges fundus images, then diagnosis can be performed, but it consumes a lot of time and effort
Solution Approach 1:
The patent replaces the manual observation and judgment process with an automated image analysis system. The system uses machine learning models to automatically extract optic disc and optic cup images, calculate cup-disc ratio, and determine glaucoma status, eliminating the time-consuming manual measurement process while significantly improving diagnosis efficiency.
Solution Approach 2:
The system performs preliminary automated extraction and processing of optic disc and optic cup images before the final diagnosis is made. By pre-processing the images and calculating intermediate parameters automatically, the system reduces the time required for the overall diagnostic process while maintaining accuracy.
3Measurement precision
If automated image analysis is introduced, then objectivity and accuracy are improved, but the system complexity increases with multiple machine learning models
Solution Approach 1:
The patent segments the diagnostic system into multiple specialized machine learning models, each responsible for a specific task: one model extracts optic disc images, another extracts optic cup images, and a third model determines glaucoma status. This segmentation allows each model to be optimized for its specific function while working together to achieve comprehensive and objective diagnosis, managing system complexity through functional decomposition.
Data Source
Figure 1
Figure 2~3
Figure 4~6
AI summary
The present disclosure provides a glaucoma image recognition method, device and screening system. The method includes: acquiring a fundus image; extracting a local image from the fundus image, the local image comprising an optic disc and a fundus background; obtaining an optic disc image and an optic cup image according to the fundus image or the local image; obtaining a disc rim image according to the optic disc image and the optic cup image; and determining whether the fundus image is classified as glaucoma according to the disc rim image, the local image and the fundus image.