Deep-Learning Ocular Landmark Measurement With Post-Hoc Refinement

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Traditional methods for measuring landmark dimensions in the human eye, such as anterior chamber depth and lens thickness, are highly dependent on the surgeon's skill set and suffer from inaccuracies, low speed, and poor repeatability.

Innovation Solution

Employing a deep-learning convolutional neural network (CNN) to automatically detect landmark features in eye images, followed by post-hoc refinement using classical image processing techniques to enhance accuracy and reduce noise.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional manual methods are used to measure landmark dimensions in the human eye, then the measurements can be obtained, but the accuracy, speed, and repeatability are poor and highly dependent on surgeon skill

Engineering Contradiction:
Improvemeasurement accuracyVSAvoiddependence on surgeon skill
Core Design Contradiction:
Measurement precisionVSEase of operation

Solution Approach 1:

The system performs automatic landmark detection and measurement calculations without requiring manual intervention or expert skill. The deep learning model independently identifies anatomical landmarks and computes dimensions, making the measurement process self-sufficient and independent of surgeon expertise.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces manual mechanical measurement methods with an automated computational system. Instead of surgeons manually measuring landmarks, a deep learning-based image processing system automatically detects features and calculates dimensions, substituting human mechanical operations with algorithmic processing.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Productivity

If traditional manual methods are used to measure landmark dimensions in the human eye, then the measurements can be obtained, but the speed of measurement is low

Engineering Contradiction:
Improvemeasurement speedVSAvoidtime for manual measurement
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The deep learning model processes images continuously and automatically without interruption. The system maintains continuous useful action by automatically detecting landmarks and calculating measurements in sequence without requiring pauses for manual intervention, thereby increasing productivity and reducing time loss.

Inventive Principle:
Principle #20Continuity of useful action

3Reliability

If traditional manual methods are used to measure landmark dimensions in the human eye, then the measurements can be obtained, but the repeatability is poor

Engineering Contradiction:
Improvemeasurement repeatabilityVSAvoidsurgeon skill dependency
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

The system performs automatic landmark detection and measurement calculations without requiring manual intervention or expert skill. The deep learning model independently identifies anatomical landmarks and computes dimensions, making the measurement process self-sufficient and independent of surgeon expertise.

Inventive Principle:
Principle #25Self-service

4Measurement precision

If deep-learning CNN is used to automatically detect landmark features, then accuracy and speed are improved, but the system complexity increases

Engineering Contradiction:
Improvelandmark detection accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces a post-hoc processing routine as an intermediary between the deep learning model and the final measurement output. This intermediate step refines the raw predictions from the CNN, corrects errors, and produces more accurate measurements, thereby managing system complexity through modular processing stages.

Inventive Principle:
Principle #24Intermediary (Mediator)

5Productivity

If deep-learning CNN is used to automatically detect landmark features, then measurement speed is improved, but noise in dimensional estimates increases

Engineering Contradiction:
Improvemeasurement speedVSAvoidnoise in dimensional estimates
Core Design Contradiction:
ProductivityVSObject-generated harmful factors

Solution Approach 1:

The patent introduces a post-hoc processing routine as an intermediary between the deep learning model and the final measurement output. This intermediate step refines the raw predictions from the CNN, corrects errors, and produces more accurate measurements, thereby managing system complexity through modular processing stages.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The post-hoc processing routine converts the noisy raw predictions from the deep learning model into accurate final measurements. By treating the initial noisy output as a starting point for refinement rather than a final result, the system transforms the harmful noise into an opportunity for iterative improvement and correction.

Inventive Principle:
Principle #22Blessing in disguise (Convert harm into benefit)

Data Source

PatentEP4244831B1Biometric ocular measurements using deep learning
Publication Date: 2026.03.04 ALCON INC
  • EP4244831B1 patent drawingFigure 1~2
  • EP4244831B1 patent drawingFigure 3
  • EP4244831B1 patent drawingFigure 4

AI summary

A method for estimating biometric landmark dimensional measurements of a human eye includes, in a possible embodiment, receiving one or more images of the human eye via a host computer. In response to receiving the one or more images, the method includes generating a preliminary set of landmark point locations in the one or more images via the host computer using a deep-learning algorithm, and then refining the preliminary set of landmark point locations using a post-hoc processing routine of the host computer to thereby generate a final set of estimated landmark point locations. Additionally, the biometric landmark dimensional measurements are automatically generated via the host computer using the final set of estimated landmark point locations. A data set is then output that is inclusive of the set of estimated landmark point locations. A host computer that executes instructions from memory to perform the method.