Retinal Layer Measurement Using Curvature-Aligned OCT Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods struggle to accurately measure the thickness of the retinal nerve fiber layer in OCT images due to large changes in curvature and deformations caused by conditions like high myopia and ophthalmic diseases, leading to difficulties in identifying layer boundaries and requiring extensive efforts to enhance accuracy.
Innovation Solution
A method involving the detection of a reference boundary line, alignment of OCT images to a baseline, and use of a deep neural network for predicting retinal layer regions, followed by calculating and restoring boundary lines using graph theory-based optimization, effectively addressing curvature issues and enhancing segmentation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional brightness gradient methods are used to detect layer boundaries, then the measurement process is simple, but measurement precision deteriorates in images with large curvature changes
Solution Approach 1:
The patent applies image preprocessing steps including noise filtering and curvature correction before boundary detection. The method pre-processes the OCT image to reduce the impact of large curvatures and noise, creating an optimized input for subsequent boundary detection algorithms, thereby improving measurement precision without significantly increasing overall complexity
Solution Approach 2:
The patent introduces intermediate processing steps including curvature calculation and correction maps that act as mediators between the raw image and final boundary detection. These intermediate representations help transform the complex problem of detecting boundaries in highly curved regions into a more manageable process with improved accuracy
2Measurement precision
If multiple preconditions and rules are set to enhance boundary detection accuracy, then measurement precision improves, but device complexity increases
Solution Approach 1:
The patent dynamically adjusts detection parameters such as gradient thresholds and search window sizes based on local image characteristics like curvature and signal-to-noise ratio. This adaptive parameter adjustment allows the algorithm to maintain high precision across different retinal regions without requiring a complex fixed rule set for every possible scenario
Solution Approach 2:
The method employs dynamic adaptation where detection parameters and search strategies are adjusted in real-time based on local image features. The algorithm adapts its behavior to local curvature changes and noise levels, reducing the need for extensive pre-configured rules while maintaining high detection accuracy
3Measurement precision
If image preprocessing and alignment steps are added to handle curvature changes, then measurement precision improves, but processing time increases
Solution Approach 1:
The patent divides the retinal image into multiple regions based on curvature characteristics and processes each region with appropriate detection parameters. This segmentation allows parallel processing of different image regions and avoids applying computationally intensive preprocessing to all areas uniformly, thereby reducing overall processing time while maintaining precision where needed
Data Source
AI summary
A method of measuring a retinal layer includes obtaining an OCT layer image of a retina, detecting a reference boundary line indicating a retinal layer in the obtained OCT image, obtaining an aligned OCT image by aligning a vertical position of each column of the OCT image so that the detected reference boundary line becomes a baseline, predicting retinal layer regions from the aligned OCT image, calculating boundary lines between the predicted retinal layer regions, and restoring the calculated boundary lines to positions of the boundary lines of the retinal layer of the original OCT image by aligning the vertical positions of the calculated boundary lines of the retinal layer for each column so that the baseline becomes the reference boundary line again.