Fundus Image Neural Networks for Non-Invasive Heart Disease Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for heart disease diagnosis lack effective non-invasive techniques that utilize fundus images to provide comprehensive diagnostic information.
Innovation Solution
A system and method utilizing an artificial neural network model trained on fundus images to assist in heart disease diagnosis, capable of detecting various heart diseases and abnormalities using deep learning techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If fundus images are used for heart disease diagnosis, then non-invasive observation of blood vessels is achieved, but diagnostic information completeness is insufficient
Solution Approach 1:
The fundus image analysis system is designed to perform multiple diagnostic functions simultaneously - detecting cardiovascular diseases, diabetic retinopathy, hypertensive retinopathy, and other systemic conditions. By training the neural network model on diverse disease patterns and clinical data, a single fundus examination serves multiple diagnostic purposes, maximizing the information extracted from non-invasive imaging.
Solution Approach 2:
The patent introduces an artificial intelligence neural network model as an intermediary between the fundus image and clinical diagnosis. This AI intermediary processes the visual information from fundus images, extracts subtle vascular patterns and abnormalities, and translates them into actionable diagnostic insights, thereby bridging the gap between non-invasive imaging and comprehensive diagnostic information.
2Measurement precision
If deep learning technology is applied to fundus image analysis, then diagnostic accuracy is improved, but system complexity increases
Solution Approach 1:
The patent extracts and isolates the core diagnostic functionality into a specialized neural network model trained specifically for fundus image analysis. By separating the AI diagnostic engine from the imaging hardware and clinical workflow, the system achieves high diagnostic accuracy through focused model training while managing complexity through modular architecture - the AI model can be updated and improved independently without redesigning the entire system.
3Adaptability or versatility
If multiple disease types are detected using fundus images, then diagnostic versatility is enhanced, but measurement precision for individual diseases may be reduced
Solution Approach 1:
The neural network model is segmented into multiple specialized detection pathways or modules, each optimized for specific disease types (e.g., cardiovascular disease detection module, diabetic retinopathy module, hypertensive retinopathy module). This segmentation allows the system to maintain high precision for each individual disease while collectively providing versatile multi-disease detection capabilities through the integrated model architecture.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to a method of assisting in diagnosis of a target heart disease using a fundus image, the method including: obtaining a target fundus image which is obtained by imaging a fundus of a testee; on the basis of the target fundus image, obtaining heart disease diagnosis assistance information of the testee according to the target fundus image, via a heart disease diagnosis assistance neural network model which obtains diagnosis assistance information that is used for diagnosis of the target heart disease according to the fundus image; and outputting the heart disease diagnosis assistance information of the testee.