Deep Learning Network for OCTA Image Analysis in AMD Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for diagnosing age-related macular degeneration (AMD) are time-consuming and lack objective standards, particularly in monitoring neovascularization using non-vascular biomarkers, and deep learning models face challenges in interpreting vascular morphology from OCTA images due to the need for large annotated databases and understanding complex network interactions.
Innovation Solution
A computer-implemented method combining optical coherence tomography angiography (OCTA) images with deep learning (DL) to classify AMD, involving pre-processing of OCTA images, inputting them into a trained DL network, and generating diagnostic results including neovascularization presence, location, and probability of AMD, using a customized convolutional neural network architecture to analyze multiple layers and extract biomarkers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning models are used to analyze OCTA images for AMD diagnosis, then diagnostic accuracy and automation are improved, but the need for large annotated databases and complex network interpretation increases device complexity and data requirements
Solution Approach 1:
The deep learning model is segmented into specialized components: a main OCTA analysis network for vascular morphology detection, and a separate fundus image analysis network for disease pattern recognition. This segmentation allows each network to be optimized for specific tasks, reducing the overall complexity while maintaining high diagnostic accuracy through coordinated analysis of multiple image types
Solution Approach 2:
Fundus images serve as an intermediary element that bridges the gap between complex OCTA vascular data and clinical diagnosis. The fundus image analysis network processes these images to provide complementary disease pattern information that helps interpret the OCTA findings, making the overall system more interpretable and clinically useful without requiring the OCTA network to handle all diagnostic complexity alone
2Measurement precision
If specialists manually monitor neovascularization changes from surrogate biomarkers, then diagnostic thoroughness is improved, but time consumption and lack of objective standards increase
Solution Approach 1:
The deep learning system performs self-service by automatically analyzing OCTA images to detect neovascularization changes and generate diagnostic assessments without requiring specialist intervention for each monitoring step. The model consistently applies objective diagnostic criteria across all cases, eliminating variability in manual assessment while maintaining thorough monitoring of vascular changes
Solution Approach 2:
The system transforms subjective specialist assessment into objective quantitative parameters by measuring specific vascular morphological features from OCTA images. This parameter-based approach provides consistent, reproducible diagnostic criteria for neovascularization detection and monitoring, replacing time-consuming manual evaluation with automated quantitative analysis
Data Source
AI summary
The present disclosure pertains to a method for diagnosing AMD comprising receiving OCTA image of a subject, pre-processing the OCTA image to obtain image data, inputting the image data to a trained deep learning (DL) network, generating using the trained DL network an output that characterizes the health of the subject with respect to AMD, and generating a diagnostic result based on the output.