Tooth Segmentation Using Geometric Primitives in Dental CBCT
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current tooth segmentation methods in dental imaging face challenges with over-segmentation and false positives, particularly when teeth are in close proximity, as they struggle to accurately differentiate foreground from background areas in volume images.
Innovation Solution
A method that involves a user-assisted approach using geometric primitives to define boundary points, forming foreground and background seed curves, and applying segmentation algorithms to CBCT images, which includes unfolding the dental arch into panoramic views to facilitate accurate segmentation by integrating human expertise with computational processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automatic segmentation algorithms are used to process dental volume images, then processing speed and productivity are improved, but segmentation accuracy deteriorates due to over-segmentation and false positives when teeth are in close proximity
Solution Approach 1:
The patent introduces geometric primitives (lines, circles, arcs) as intermediary tools that bridge automatic algorithms and manual interaction. These primitives serve as mediators by providing structured guidance for boundary detection, allowing the system to maintain automated processing efficiency while achieving manual-level accuracy in distinguishing foreground from background regions, particularly for closely spaced teeth
Solution Approach 2:
The patent divides the segmentation process into distinct stages: initial automatic processing, geometric primitive-based boundary identification, and refined segmentation. This multi-stage segmentation approach allows the system to first quickly process the volume image automatically, then apply geometric constraints to correct potential errors, and finally achieve accurate segmentation without sacrificing overall processing speed
2Measurement precision
If manual boundary identification methods are used to improve segmentation accuracy, then measurement precision is improved, but device complexity and ease of operation worsen due to increased user interaction requirements
Solution Approach 1:
The patent transforms the interaction paradigm by changing parameters from freehand drawing to constrained geometric primitive selection. Users select from predefined geometric forms (lines, circles, arcs) with controlled degrees of freedom, which simplifies the interaction model while maintaining or improving boundary identification accuracy compared to completely manual methods
Solution Approach 2:
The system provides self-service by automatically generating and fitting geometric primitives to detected boundaries. Once a user initiates the process or provides minimal input, the system autonomously fits appropriate geometric forms to tooth boundaries, reducing the burden on users while maintaining high segmentation accuracy
3Measurement precision
If geometric primitives are used to define boundary points, then segmentation accuracy is improved by reducing over-segmentation, but ease of operation deteriorates due to the need for user assistance
Solution Approach 1:
Geometric primitives serve as intermediaries that translate complex boundary detection tasks into simple selection operations. Instead of requiring users to manually trace complex tooth boundaries, they select from predefined geometric forms that automatically fit to the data, maintaining ease of operation while achieving high boundary identification accuracy
Solution Approach 2:
The system uses geometric primitives as simplified copies or representations of actual tooth boundaries. These geometric forms capture the essential boundary characteristics without requiring precise manual tracing, making the operation easier while maintaining sufficient accuracy for segmentation purposes
Data Source
AI summary
A method for segmenting a feature of interest from a volume image acquires image data elements from the image of a subject. One or more boundary points along a boundary of the feature of interest are identified according to one or more geometric primitives with reference to the displayed view. A foreground seed curve is defined according to the one or more identified boundary points. A background field array that lies outside of, and is spaced from, the foreground seed curve by a predetermined distance, is defined. Segmentation is applied to the volume image according to foreground values obtained according to image data elements that are spatially bounded on or within the foreground seed curve and according to background field array values to create a segmented feature of interest.