Automated Geographic Atrophy Lesion Segmentation for Progression Prediction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods for evaluating geographic atrophy (GA) lesions in age-related macular degeneration are time-consuming and prone to inter- and intra-observer variability, lacking accuracy in segmentation and progression prediction.

Innovation Solution

Employing a deep learning system, particularly a convolutional neural network (CNN) or U-Net model, for fully automated segmentation and feature evaluation of GA lesions, using fundus autofluorescence and optical coherence tomography images to generate precise segmentation masks and predict GA progression based on shape and textural features.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If manual evaluation methods are used for GA lesions, then flexibility and adaptability are maintained, but time consumption and observer variability increase

Engineering Contradiction:

Improvemanual evaluation flexibilityVSAvoidevaluation time

Core Design Contradiction:

Ease of operationVSLoss of time

Solution Approach 1:

The patent replaces manual mechanical evaluation with an automated computer-based system that processes fundus autofluorescence images. The system uses image analysis algorithms to automatically segment GA lesions, extract shape and textural features, and predict progression, eliminating the need for manual observation and measurement while significantly reducing evaluation time.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The evaluation system performs self-assessment by automatically analyzing its own input images without requiring external manual intervention. The computer-based system independently completes all steps from image processing to progression prediction, making the evaluation process autonomous and eliminating observer-dependent variability.

Inventive Principle:

Principle #25Self-service

2Measurement precision

If manual segmentation is used for GA lesions, then detailed analysis is possible, but inter-observer and intra-observer variability increase

Engineering Contradiction:

Improvesegmentation accuracyVSAvoidobserver consistency

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The patent replaces manual segmentation with an automated image analysis system that consistently applies the same algorithms to all images. This computer-based approach eliminates human variability in boundary detection and lesion delineation, providing both high precision and reliable reproducible results across different observers and time points.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Productivity

If automated segmentation is implemented, then time consumption is reduced, but segmentation accuracy may deteriorate

Engineering Contradiction:

Improveevaluation speedVSAvoidsegmentation accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent employs sophisticated computer-based image analysis algorithms that automatically segment GA lesions from fundus autofluorescence images. These algorithms maintain high segmentation accuracy by using advanced image processing techniques while simultaneously providing rapid evaluation, thus achieving both high productivity and measurement precision without the trade-off present in simpler automated systems.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

4Measurement precision

If comprehensive feature extraction is performed, then prediction accuracy is improved, but computational complexity increases

Engineering Contradiction:

Improveprogression prediction accuracyVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the complex feature extraction process into distinct modules: shape feature extraction (area, perimeter, circularity) and textural feature extraction (autofluorescence patterns). This segmentation of the analysis process allows comprehensive feature extraction for accurate prediction while managing computational complexity through organized, modular processing steps.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent extracts only the most relevant features from the images - specifically shape features and textural features from fundus autofluorescence. By selectively extracting only the necessary information rather than processing all possible image data, the system achieves high prediction accuracy while keeping computational complexity manageable.

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS12430756B2Prediction of geographic-atrophy progression using segmentation and feature evaluation

Publication Date: 2025.09.30 GENENTECH INC

AI summary

A method, system, and computer program product for evaluating a geographic atrophy lesion. An image of a geographic atrophy (GA) lesion is received. A first set of values is determined for a set of shape features using the image. A second set of values is determined for a set of textural features using the image. GA progression for the GA lesion is predicted using the first set of values and the second set of values.