Fundus Image CNN Screening for Quantitative Glaucoma Likelihood
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current primary open-angle glaucoma screening methods are unreliable, subjective, and expensive, leading to a significant portion of cases going undiagnosed due to low sensitivity and specificity, hindering effective screening programs.
Innovation Solution
A system utilizing a convolutional neural network to analyze fundus images, applying variations to optic nerve head regions, generating and combining likelihood estimates to provide a quantitative assessment of primary open-angle glaucoma likelihood.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current screening methods are used, then the screening process can be performed, but the sensitivity and specificity are low leading to many undiagnosed cases
Solution Approach 1:
The patent replaces subjective manual screening methods with an automated convolutional neural network system that processes fundus images to detect glaucoma signs. This substitution of mechanical/subjective evaluation with computational automation enables consistent, objective assessment that improves both sensitivity and specificity by eliminating human observer variability and enhancing detection precision through deep learning algorithms.
Solution Approach 2:
The patent transforms the screening approach by changing from binary diagnostic categories to a continuous likelihood estimate scale. Instead of simple positive/negative diagnoses, the system outputs a probability score between 0 and 1 representing glaucoma likelihood, enabling more nuanced and precise measurement of diagnostic accuracy while improving sensitivity and specificity through probabilistic assessment.
2Productivity
If large-scale screening programs are adopted, then more people can be screened, but the costs and potential harm outweigh the benefits due to low accuracy
Solution Approach 1:
The patent enables large-scale screening by replacing labor-intensive manual evaluation with automated image processing. The convolutional neural network can rapidly analyze numerous fundus images without human intervention, dramatically increasing screening capacity and productivity while maintaining high reliability through consistent algorithmic performance, thereby making large-scale screening economically viable and effective.
3Ease of manufacture
If manual screening methods are used, then the process is simple to implement, but the subjectivity and cost reduce effectiveness
Solution Approach 1:
The patent maintains ease of implementation by requiring only fundus image acquisition and automated processing, eliminating the need for complex manual evaluation protocols. The system's simplicity in operation—input image, receive likelihood estimate—combines with automated consistency to provide both ease of manufacture and high diagnostic reliability, resolving the contradiction between simplicity and consistency.
Data Source
AI summary
This application describes a system to estimate the likelihood of glaucoma from fundus images. The system may include receiving an optic nerve head region of an original image. A variation to the optic nerve head region are applied. A first primary open angle glaucoma likelihood estimate based on the first varied optic nerve head region is generated using a trained convolutional neural network. A second primary open angle glaucoma likelihood estimate based on the second varied optic nerve head region is generated using a trained convolutional neural network. The first primary open angle glaucoma likelihood estimate and the second primary open angle glaucoma likelihood estimate are combined into a final primary open angle glaucoma likelihood estimate. The final primary open angle glaucoma likelihood estimate is presented to a user.