Automated Airway Cross-Section Visualization Without Pre-Computed Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for quantitative airway analysis using high-resolution CT images are limited by the need for pre-computed bronchial tree segmentations, which are time-consuming and often fail to include distal airway branches, making it difficult to visualize and measure airways not aligned with the axial plane.
Innovation Solution
An automated method that allows users to select any airway in a 3D lung image, compute the principle axes, and display a perpendicular 2D cross-section instantly, enabling visualization and measurement without pre-computed segmentations, using techniques like adaptive thresholding and eigenvector computation for 3D segmentation and interpolation for cross-section generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If pre-computed bronchial tree segmentations are used to generate cross-sections, then automated airway evaluation can be performed, but the process becomes time-consuming and fails to include distal airway branches
Solution Approach 1:
The patent extracts the essential function of generating perpendicular cross-sections from the complex pre-computed bronchial tree segmentation process. Instead of requiring complete automated segmentation of the entire bronchial tree, the system extracts and processes only the specific airway segment of interest that the user selects, thereby achieving automated evaluation without the time-consuming complete segmentation.
Solution Approach 2:
The patent divides the bronchial tree segmentation into two parts: (1) the complete automated segmentation for reference, and (2) the user-selectable specific airway segments for detailed analysis. This allows the system to leverage automated segmentation capabilities while enabling flexible, on-demand analysis of specific regions without requiring complete segmentation processing time.
2Measurement precision
If manual or semi-automatic bronchial tree extraction is performed, then orientation information can be obtained, but the process is time-consuming and does not reach distal airway branches
Solution Approach 1:
The patent enables the system to automatically generate orientation information and perpendicular cross-sections without requiring manual or semi-automatic user intervention. The user simply selects an airway of interest, and the system automatically computes the orientation, generates the cross-section, and enables measurement, thereby eliminating the time-consuming manual extraction process while maintaining measurement precision.
3Adaptability or versatility
If complete bronchial tree segmentation is performed, then all airway branches can be visualized, but distal airway branches remain inaccessible for measurement
Solution Approach 1:
The patent transforms the static complete segmentation into a dynamic, on-demand system. Instead of having a fixed complete segmentation that may miss distal branches, the system dynamically generates segmentation and cross-sections for specific airways of interest, including distal branches, based on user selection. This allows adaptive visualization and measurement of any airway region regardless of its position in the bronchial tree.
4Measurement precision
If manual specification of airway orientation is performed, then accurate cross-sections can be obtained, but user workload increases and measurement variability increases
Solution Approach 1:
The patent implements automated orientation computation that eliminates the need for manual specification of airway orientation. The system automatically calculates the perpendicular cross-sections based on the selected airway, thereby maintaining accurate measurements while significantly reducing user workload and eliminating measurement variability caused by manual intervention.
Data Source
AI summary
A method including displaying a three-dimensional (3D) image of a lung, receiving a selection of an airway of the lung and displaying a two-dimensional (2D) cross-section image of the airway perpendicular to the airway's long axis, wherein the display of the 2D cross-section image occurs almost immediately after the selection of the airway is received.