Vessel Extraction Using Bone Masking and Adaptive Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for extracting vasculature from angiography images face challenges such as excessive extraction of bones and other tissues as vessels, and incomplete extraction of small vessels due to overlapping pixel or voxel values.
Innovation Solution
The method involves obtaining a first image by processing the angiography image using a second threshold and Laplacian enhancement, or by extracting bone tissue using a third threshold. This is followed by identifying vessel seed groups, determining vessel thresholds, and performing adaptive thresholding operations to refine the vessel segmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional vessel extraction methods are used, then extraction speed is maintained, but bones and other tissues are extracted as vessels resulting in excessive extraction
Solution Approach 1:
The patent divides the vessel extraction process into multiple stages: initial thresholding to obtain candidate vessels, connected component analysis to identify and remove bone regions, and refined thresholding to extract true vessels. This multi-stage segmentation approach separates the extraction of different tissue types, preventing bones from being mistakenly identified as vessels while ensuring complete vessel detection.
2Measurement precision
If traditional vessel extraction methods are used, then processing simplicity is maintained, but small vessels are removed as non-vessel tissue resulting in incomplete extraction
Solution Approach 1:
The patent performs preliminary bone removal by identifying and eliminating bone-connected components before final vessel extraction. By removing bone regions in advance, the subsequent thresholding operation can focus solely on vessel structures without mistakenly classifying bone tissue as vessels, thereby ensuring small vessels are not removed as non-vessel tissue.
Solution Approach 2:
The patent introduces an intermediary step of connected component analysis and bone mask creation between initial thresholding and final vessel extraction. This intermediary process acts as a mediator that separates bone structures from vessel structures, allowing the final extraction to achieve high completeness without requiring overly complex direct segmentation methods.
3Measurement precision
If threshold values are lowered to capture small vessels, then vessel detection sensitivity improves, but bones and other tissues are more likely to be extracted as vessels
Solution Approach 1:
The patent applies different processing strategies to different regions of the image. After initial thresholding, connected component analysis is applied locally to identify bone regions based on their spatial relationships and intensity characteristics. This local analysis allows the use of lower global thresholds to capture small vessels while preventing false extraction of bone and other non-vessel tissues through region-specific filtering.
Data Source
Figure 1
Figure 2~3
Figure 4-A~4-B
AI summary
A method and system for extracting vasculature. The method may include obtaining an image including a part representing a vessel, and obtaining a first image by extracting bone tissue in the image. The method may also include obtaining a second image by performing a region growing operation on the image. The method may further include generating a subtraction image based on the first image and the second image. The subtraction image may include at least a portion of the part representing the vessel in the image. The method may also include obtaining at least one vessel seed group based on a first condition and the subtraction image, and determining at least one vessel threshold based on at least one boundary distance field of the subtraction image. The method may further include obtaining a target vessel image by performing, based on the vessel seed group and the at least one vessel threshold, a region growing operation on the image.