Neural Network Model for Multi-Disease Diagnosis from Eye Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current diagnostic methods using fundus images are limited in their ability to assist in the detection of systemic diseases beyond eye-related conditions, and there is a need for a more comprehensive and accurate system for analyzing eye images to provide multi-faceted diagnosis assistance.
Innovation Solution
A neural network model is developed that includes multiple layers to process eye images, utilizing a combination of common and individual portions for feature extraction and diagnosis assistance, trained on distinct data sets to provide comprehensive diagnosis assistance information for both eye and systemic diseases.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a single neural network model is used for diagnosis assistance, then the device complexity is low, but the versatility and comprehensiveness of diagnosis assistance information is limited
Solution Approach 1:
The patent divides the neural network model into multiple specialized sub-models, where each sub-model is trained on specific training data for particular diseases or diagnostic tasks. This segmentation allows each model to specialize in specific diagnostic aspects while collectively providing comprehensive diagnosis assistance, resolving the contradiction between versatility and complexity.
Solution Approach 2:
The patent creates a multi-functional diagnostic system where multiple neural network models work together to provide various types of diagnosis assistance information. Each model contributes to different diagnostic functions, enabling the system to handle diverse diagnostic tasks (eye diseases, systemic diseases) while maintaining manageable complexity through modular architecture.
2Measurement precision
If multiple neural network models are used to provide comprehensive diagnosis assistance information, then the versatility and accuracy of diagnosis is improved, but the device complexity and computational requirements increase
Solution Approach 1:
The patent segments the diagnostic task into multiple specialized neural network models, each trained on specific training data for particular diseases or diagnostic aspects. This segmentation improves accuracy by allowing each model to specialize in specific diagnostic tasks while managing complexity through modular, independent model structures that can be processed separately.
Data Source
AI summary
An aspect of the present invention relates to a diagnostic assistance device for acquiring diagnostic assistance information by using a neural network model and based on an eye image, the diagnostic assistance device comprising: an eye image acquisition unit for acquiring a target eye image; and a processing unit for acquiring diagnostic assistance information by using a neural network model learned to acquire diagnostic assistance information and based on the target eye image. The neural network model includes first diagnostic assistance neural network model and second diagnostic assistance neural network model for acquiring second diagnostic assistance information. The first diagnostic assistance neural network model includes first common portion for acquiring first feature set and first individual portion for acquiring first diagnostic assistance information, and the second diagnostic assistance neural network model includes first common portion for acquiring first feature set and second individual portion for acquiring second diagnostic assistance information.