Color Fundus Photograph Analysis for Glaucoma Risk Stratification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting glaucoma onset and progression rely on multiple test modalities that are not readily available in primary healthcare settings, making timely detection challenging.
Innovation Solution
A computer-implemented method using machine-learning classifiers trained on color fundus photographs to diagnose glaucoma and predict its incidence or progression, employing segmentation of anatomical structures like retinal vessels, macula, and optic cup.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple test modalities (IOP measurement, visual field tests) are used to predict glaucoma incidence and progression, then prediction accuracy is improved, but device complexity and ease of operation deteriorate due to unavailability in primary healthcare settings
Solution Approach 1:
The patent extracts and focuses on a single, readily available test modality (color fundus photograph) from the multiple required test modalities (IOP measurement, visual field tests, CFP). By isolating CFP as the sole input, the system eliminates the complexity of coordinating multiple specialized devices while maintaining predictive capability through AI-based analysis of the extracted anatomical structures (optic disc, optic cup, macula, retinal vessels).
Solution Approach 2:
The patent makes the color fundus photograph serve multiple functions: it is used for both diagnosis of existing glaucoma and prediction of future glaucoma incidence and progression. This multi-functionality eliminates the need for separate specialized tests, allowing a single widely available device to perform what previously required multiple specialized modalities.
2Measurement precision
If multiple test modalities are used for glaucoma prediction, then prediction accuracy is improved, but ease of operation worsens due to requirements for specialized equipment and trained personnel
Solution Approach 1:
The system extracts only the essential information needed for prediction from the complex array of test modalities, focusing solely on anatomical structures visible in color fundus photographs. This extraction approach maintains prediction accuracy by concentrating on the most relevant features (optic disc, optic cup, macula, retinal vessels) while eliminating the operational burden of performing and coordinating multiple specialized tests.
3Ease of operation
If only color fundus photographs are used for glaucoma prediction, then ease of operation and availability are improved, but measurement precision deteriorates compared to multiple test modalities
Solution Approach 1:
The patent replaces manual clinical assessment and interpretation of multiple test modalities with an AI-based machine learning system that automatically analyzes color fundus photographs. This substitution of mechanical/manual processes with automated computational analysis enables the system to extract subtle patterns and features from CFPs that would be difficult for humans to detect, thereby maintaining high prediction accuracy while using only the simple, accessible CFP modality.
Solution Approach 2:
The patent transforms the raw color fundus photograph data into meaningful predictive parameters through AI processing. By changing the parameters from simple image data to extracted anatomical measurements and AI-derived risk scores, the system enhances the informational content of the CFP, enabling accurate prediction despite using a single, simple modality.
Data Source
AI summary
Deep learning based systems and methods for predicting and stratifying the risk of glaucoma onset and progression based on color fundus photographs (CFPs) are disclosed. The methods are clinically validated by external population cohorts wherein to apply a machine-learning classifier having been trained using a dataset of CFPs of a longitudinal patient cohort regarding glaucoma development of each of the patients in the cohort over a period of time (e.g., over the course of a few vears), to predict a likelihood of glaucoma incidence or progression for the patient in the future (e.g., over a similar period of time of several years).