Wavefront Vision Assessment Using Zernike Features and Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional eye examination procedures only assess lower-order aberrations, failing to account for higher-order aberrations that significantly impact vision quality, and lack comprehensive vision correction solutions.
Innovation Solution
A system utilizing machine learning models, specifically neural networks and support vector regression, processes wavefront aberration data expressed in Zernike polynomials to generate vision correction factors for refractive surgery, spectacles, and contact lenses, incorporating training datasets for personalized assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional eye examination procedures are used, then the assessment process is simple and quick, but only lower-order aberrations are assessed, failing to account for higher-order aberrations that significantly impact vision quality
Solution Approach 1:
The patent segments the wavefront aberration data into distinct components using Zernike polynomials, separating lower-order aberrations (defocus, astigmatism) from higher-order aberrations (coma, trefoil, spherical aberration). This segmentation allows the system to comprehensively assess all aberration types while maintaining structured data processing through the machine learning model.
Solution Approach 2:
The patent introduces a machine learning model as an intermediary between wavefront aberration measurement and vision quality assessment. This intermediary processes the complex wavefront data and Zernike polynomial coefficients to generate accurate vision quality predictions, bridging the gap between raw measurement data and clinically meaningful outcomes without requiring direct complex mathematical transformations.
2Measurement precision
If machine learning models with multiple input factors are used, then vision correction prediction accuracy is improved, but computational complexity and data processing requirements increase
Solution Approach 1:
The patent performs preliminary processing of wavefront aberration data by expressing it as Zernike polynomials and extracting relevant coefficients before feeding them into the machine learning model. This preliminary action organizes the data in a standardized format with predefined input factors, reducing the computational burden during the actual prediction phase and enabling faster inference while maintaining high accuracy.
Solution Approach 2:
The patent transforms the raw wavefront aberration data into Zernike polynomial coefficients, changing the parameter representation from spatial domain to frequency domain. This parameter transformation simplifies the data structure and enables the machine learning model to process aberration information more efficiently, improving prediction accuracy without proportionally increasing computational complexity.
3Loss of information
If comprehensive wavefront analysis is performed, then higher-order aberrations are captured, but measurement and data processing time increase
Solution Approach 1:
The patent extracts only the most clinically relevant Zernike polynomial coefficients from the complete wavefront analysis data, selecting specific input factors (defocus, astigmatism, coma, trefoil, spherical aberration coefficients) that have the greatest impact on vision quality. This extraction approach captures essential aberration information while discarding redundant data, reducing processing time without significant loss of diagnostic value.
Solution Approach 2:
The patent implements a balanced approach by performing complete wavefront analysis to capture all aberration information, then using the machine learning model to process only the essential components needed for vision quality assessment. This partial processing approach ensures comprehensive data collection while limiting the computational burden to only the most critical parameters, optimizing the trade-off between data completeness and processing efficiency.
Data Source
AI summary
A system and method of assessing vision quality of an eye is presented, with a controller having a processor and tangible, non-transitory memory on which instructions are recorded. The controller is configured to selectively execute at least one machine learning model. Execution of the instructions by the processor causes the controller to: receive wavefront aberration data of the eye and express the wavefront aberration data as a collection of Zernike polynomials. The controller is configured to obtain a plurality of input factors based on the collection of Zernike polynomials. The plurality of input factors is fed into the at least one machine learning model, which is trained to analyze the plurality of input factors. The machine learning model generates at least one vision correction factor based in part on the plurality of input factors.