Dry Eye Classification via Tear Film Interference Fringe Color Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image classification methods struggle to accurately classify dry eye types using interference fringe images of the tear fluid layer due to the lack of distinct feature points and reliance on subjective observer expertise, leading to biased results.
Innovation Solution
An image classification method that extracts feature values from interference fringe images based on color information, including luminance averages, standard deviations, and local color variations, using machine learning to objectively classify dry eye types, particularly focusing on 'healthy person,' 'Aqueous Deficient Dry Eye,' and 'Evaporative Dry Eye' types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image classification methods are used to classify dry eye types from interference fringe images, then the classification process can be performed, but the classification accuracy is insufficient and results are biased due to subjectivity
Solution Approach 1:
The patent replaces the manual mechanical process of observer-based image classification with an automated machine learning system. The system extracts color information (RGB values) and luminance features from interference fringe images, then uses supervised learning algorithms to automatically classify dry eye types, eliminating human subjectivity and improving both accuracy and reliability.
2Loss of information
If traditional feature extraction methods focusing on edge or corner points are used, then local features can be extracted, but feature values are insufficient because interference fringe images lack distinct feature points
Solution Approach 1:
The patent changes the parameter basis for feature extraction from geometric features (edges, corners) to colorimetric parameters (RGB values, luminance, saturation). By extracting color information from all pixels rather than relying on sparse feature points, the system captures comprehensive feature data from interference fringe images that lack distinct geometric features.
Solution Approach 2:
The patent transitions from two-dimensional spatial feature extraction (edges and corners in x-y coordinates) to a three-dimensional color space analysis (RGB values plus derived luminance and saturation parameters). This dimensional expansion provides richer feature information from images that are deficient in traditional geometric features.
3Productivity
If observer-based classification methods are used, then classification can be performed using expert knowledge, but the process is time-consuming and subject to inter-observer variability
Solution Approach 1:
The patent implements a self-service classification system where the machine learning model automatically performs classification without requiring human observer intervention. The system trains on labeled data and then independently classifies new images, eliminating inter-observer variability and significantly improving both speed and consistency of classification results.
Data Source
AI summary
In classifying images by machine learning, provided are an image classification method, device, and program for classifying the image from which the feature difference is hardly detected, in particular, classifying the interference fringe image of tear fluid layer by the dry eye types. The method includes a step of acquiring a feature value from an interference fringe image of tear fluid layer for learning, a step of constructing a model for classifying an image from the feature value acquired from the interference fringe image of tear fluid layer for learning, a step of acquiring the feature value from an interference fringe image of tear fluid layer for testing, and a step of performing classification processing for classifying the interference fringe image of tear fluid layer for testing by types of dry eye using the model and the feature value acquired from the interference fringe image of tear fluid layer.