Smartphone Pupil Reflection Refractive Error Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Accurate diagnosis of refractive errors in newborns and infants is challenging due to their inability to cooperate with conventional inspection methods, which are also prone to inaccuracy and lengthy, and may lead to delayed treatment and amblyopia.
Innovation Solution
A method using a smartphone camera to capture a reflection image of the pupil, applying it to a ResNet model trained with clinical data to determine refractive error, outputting results in the form of a heat map, allowing for quick and accurate measurement without subject cooperation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If cycloplegic refraction is performed using a skiascope, then measurement precision of refractive error is improved, but inspection time is extended and subject cooperation is required
Solution Approach 1:
The patent replaces the mechanical skiascope-based cycloplegic refraction system with a smartphone camera-based photorefraction system. The smartphone camera captures images of the pupil and retina, and deep learning algorithms automatically analyze these images to determine refractive error, eliminating the need for mechanical skiascope operations and cycloplegic eye drops.
Solution Approach 2:
The system enables self-service measurement where the smartphone automatically captures images and the deep learning model automatically processes them to provide refractive error measurement. This eliminates the need for skilled inspector operation and subject cooperation, allowing independent measurement.
2Measurement precision
If cycloplegic refraction is performed by a skilled oculist, then measurement precision is improved, but device complexity and operational difficulty increase
Solution Approach 1:
The deep learning model performs automatic analysis of captured images to determine refractive error, eliminating the need for skilled inspector interpretation. The system self-processed the images and provides measurements without requiring expert knowledge or manual manipulation.
Solution Approach 2:
The patent replaces the skilled inspector's manual skiascope operation with an automated deep learning-based image analysis system. The smartphone camera captures images and the AI model automatically processes them, substituting human expertise with algorithmic analysis.
3Measurement precision
If conventional refraction methods are used on infants, then measurement precision may be improved, but ease of operation deteriorates due to lack of subject cooperation
Solution Approach 1:
The system is designed to work without subject cooperation by using photorefraction images that can be captured of the infant's eye without requiring them to understand or participate in the measurement process. The deep learning model automatically analyzes these images to provide measurements.
Solution Approach 2:
The patent replaces the interaction-based skiascope method requiring subject cooperation with a passive image capture method using smartphone camera. The system captures photorefraction images and uses AI analysis, eliminating the need for subject participation or understanding of the measurement process.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables rapid, accurate measurement of refractive errors in infants, reducing the risk of amblyopia and improving eyesight development, with high sensitivity and specificity, and is cost-effective without requiring expensive equipment.
Implementation Method 1
obtaining an image by projecting a visible ray using a flash onto the retina of a subject and capturing, using a smartphone camera, a shape and a location reflected in the pupil of the subject
Data Source
AI summary
Disclosed is a method of measuring a refractive error using a reflection image of a pupil for a visible ray, the method capable of diagnosing and self-examining a refractive error using only an image captured by a smartphone camera. The method may include obtaining an image by projecting a visible ray, using a flash, onto the retina of a subject and capturing, using a smartphone camera, a shape and location of the image reflected in the pupil of the subject, applying the image to a residual neural network (ResNet) model trained using data of an ImageNet based on clinical information into which result values of a refractive error of the subjects have been incorporated, and outputting the refractive error of the image and a measurement factor measured to determine the refractive error based on the shape and the location reflected in the pupil in a process of deriving the refractive error of the image.