Optic Disc Segmentation in OCT Retinal Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current optical coherence tomography (OCT) methods face challenges in accurately segmenting the optic nerve head (ONH) region due to its unique structural differences from neighboring retinal areas, leading to segmentation errors and exclusion of valuable diagnostic information.
Innovation Solution
A method and apparatus that utilize two-dimensional OCT images to detect and exclude the ONH cutout region from analysis, allowing for precise identification and quantification of retinal layers by processing image data to separate the region of interest from the rest of the image, using techniques such as gradient evaluation and landmark detection to refine the boundaries and volume estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If traditional OCT methods are used to segment the optic nerve head region, then the segmentation process can be performed automatically, but segmentation errors occur due to unique structural differences from neighboring retinal areas
Solution Approach 1:
The patent divides the retinal image into multiple distinct regions: the optic nerve head (ONH) region, the peripapillary region, and the remaining retina. By segmenting the image into these separate zones and applying different analysis methods to each, the system handles the unique structural characteristics of the ONH without allowing them to compromise the overall segmentation accuracy of the entire image.
Solution Approach 2:
The patent extracts and excludes the ONH cutout region from the automated layer segmentation process. By removing this problematic region from the analysis, the system prevents the unique structural features of the ONH from causing segmentation errors in the surrounding retinal layers, thereby improving overall measurement precision.
2Area of stationary object
If the ONH region is included in automated layer segmentation, then the process covers the entire retina, but segmentation errors and deviations in layer boundaries occur adjacent to the ONH
Solution Approach 1:
The patent identifies and extracts the ONH cutout region from the retinal image, then excludes this region from automated layer segmentation. This extraction approach maintains coverage of the entire retina for diagnostic purposes while preventing the ONH's unique structure from causing segmentation errors in adjacent areas.
Solution Approach 2:
The patent applies different processing qualities to different regions: the ONH region is excluded from automated segmentation to maintain local accuracy, while the surrounding retinal regions receive full automated analysis. This local quality approach ensures that each region is processed according to its specific structural characteristics.
3Measurement precision
If the ONH region is excluded from analysis to avoid segmentation errors, then layer segmentation accuracy improves, but valuable diagnostic information from the ONH region is lost
Solution Approach 1:
The patent segments the retinal image to identify the ONH cutout region, then excludes only this specific region from automated layer segmentation while preserving it for separate diagnostic analysis. This segmentation approach maintains layer boundary accuracy in the surrounding retina while retaining valuable ONH diagnostic information for independent evaluation.
Data Source
Figure 1
Figure 2~4
Figure 3
AI summary
Provided is a method of processing image data and detecting a region of an image represented by the image data to be excluded from an analysis of the image. According to the method, image data captured by a medical modality is received. An evaluation of a portion of the image data representing a two-dimensional view of a subject appearing in the image is conducted to locate, in the two-dimensional view, the region to be excluded from the analysis of the image. A feature pertinent to the analysis appearing in a remaining portion of the image, that is outside of the region to be excluded from the analysis located by the evaluation, is detected.