Optic Nerve Head Classification via Deep Neural Network Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting changes in the optic nerve head, particularly for glaucoma diagnosis, are limited by reliance on cup-disc ratio measurements, which are not definitive for axonal loss and are challenging to apply accurately and universally, especially in non-specialized settings, and existing biometric technologies face issues with robustness and accessibility.
Innovation Solution
A computer-implemented method using deep neural networks for automatic recognition and classification of optic nerve head images, segmenting the optic nerve head and analyzing blood vessel relationships to detect changes independent of cup-disc ratio, enabling early detection of glaucoma and age assessment, and utilizing machine learning for biometric identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If cup-disc ratio measurement is used for glaucoma detection, then the method is simple to implement, but the measurement precision and reliability are insufficient
Solution Approach 1:
The optic nerve head image is segmented into multiple regions including the optic disc, neuroretinal rim, and optic cup using deep neural networks. This segmentation enables precise measurement of individual structures and their relationships, moving beyond the single cup-disc ratio metric to multiple independent measurements that collectively provide more accurate detection of axonal loss.
Solution Approach 2:
The invention transitions from two-dimensional cup-disc ratio measurement to three-dimensional volumetric analysis of the optic nerve head structures. By calculating volumes of the optic disc, neuroretinal rim, and optic cup, the system provides more comprehensive and accurate assessment of structural changes associated with glaucoma progression.
2Measurement precision
If specialized medical equipment and skilled ophthalmologists are used for optic nerve head examination, then the diagnostic accuracy is high, but the device complexity and accessibility are poor
Solution Approach 1:
The system employs automated deep neural networks that can independently perform image segmentation, feature extraction, and glaucoma detection without requiring specialized ophthalmological expertise. The algorithm automatically identifies the optic disc, neuroretinal rim, and optic cup, and calculates their volumes and relationships, enabling non-specialized operators to achieve accurate diagnosis.
Solution Approach 2:
The invention replaces manual examination by skilled ophthalmologists using complex medical equipment with an automated computer vision system. Deep neural networks process the fundus images to extract diagnostic features and provide glaucoma assessment, substituting human expertise and mechanical examination tools with an intelligent algorithmic system that is both accurate and accessible.
3Ease of operation
If cup-disc ratio is used as the primary metric, then the measurement process is straightforward, but the reliability for definitive glaucoma diagnosis is insufficient
Solution Approach 1:
The system calculates multiple metrics simultaneously including the volumes of the optic disc, neuroretinal rim, and optic cup, as well as their spatial relationships. This multi-functional approach provides a comprehensive set of measurements that can collectively indicate glaucoma with higher reliability than any single metric, while maintaining computational efficiency through automated processing.
Data Source
Figure 1a~2a
Figure 2b~3
Figure 4~5
AI summary
Provided are a method and system for identifying and classifying the owner, age and health of an optic nerve head and its vasculature based on analysis of vector relationships of blood vessels and the neuroretinal rim within an image of the optic nerve head to each other.