Corneal OCT Edge Joining for Bowman's Layer Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing ophthalmic technologies struggle to accurately segment the locally uneven layers of the cornea in anterior segment tomographic images due to low contrast and noise in OCT images, particularly the Bowman's layer, making precise segmentation challenging.
Innovation Solution
An anterior segment analysis apparatus and method that utilizes OCT measurements to detect edges in tomographic images, performs edge joining based on specific conditions, and determines layer boundaries by selecting and joining edges to enhance segmentation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional OCT imaging is used to capture corneal layers, then the imaging process is simple and fast, but the segmentation accuracy of locally uneven layers (especially Bowman's layer) is poor due to low contrast and noise
Solution Approach 1:
The patent applies segmentation by dividing the edge detection and joining process into multiple stages: initial edge detection, first joining operation, selection of representative edges, second joining operation, and final boundary determination. This multi-stage segmentation approach improves segmentation accuracy of locally uneven corneal layers while managing processing complexity through systematic breakdown of the processing task
Solution Approach 2:
The patent implements local quality by applying different joining conditions to different regions of the corneal image. The first joining condition and second joining condition are tailored to handle specific local characteristics of corneal layers, particularly addressing the low contrast and noise issues in specific regions like Bowman's layer, thereby improving local segmentation accuracy without uniformly increasing processing complexity across the entire image
2Reliability
If edge detection is performed on low contrast OCT images, then more edges may be detected, but noise and false edges increase making accurate segmentation difficult
Solution Approach 1:
The patent applies preliminary action by performing the first joining operation before final boundary determination. This preliminary joining step connects nearby edges and fills gaps in the edge detection results, creating a more reliable edge structure that reduces the impact of noise and false edges before the final segmentation decision is made
Solution Approach 2:
The patent implements feedback through the iterative process of edge detection, joining, selection, and re-joining. The second joining operation uses information from the first joining and edge selection steps to refine the boundary determination, creating a feedback loop that continuously improves segmentation reliability while filtering out noise and false edges
Data Source
AI summary
An anterior segment analysis apparatus includes a control device configured to: obtain a tomographic image of an anterior segment including a cornea of a subject eye formed by OCT measurement; detect a plurality of edges included in the tomographic image; perform a first joining including joining between edges for each of the plurality of edges based on a first joining condition; select a first edge and a second edge from among the joined edges based on a length; performs a second joining including joining between edges with reference to each of the first edge and the second edge based on a second joining condition; and determine a boundary of a layer of the cornea of the tomographic image using the edges joined by the second joining.