3D Bump Localization for Lung Nodule Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for analyzing 3D medical images, such as CT scans, are inefficient and inaccurate in detecting irregularities like lung nodules attached to the pleura, as they rely on 2D slice analysis, leading to inconsistent segmentation results.
Innovation Solution
A 3D bump localization method using local surface differences, which involves selecting a volume of interest, growing a surface based on intensity and spatial constraints, applying filters, and determining histogram thresholds to detect irregularities, enabling more accurate segmentation of abnormalities like lung nodules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If 2D slice analysis is used to detect lung nodules, then processing time is reduced, but segmentation accuracy and consistency deteriorate
Solution Approach 1:
The patent transitions from 2D slice-based analysis to 3D volume-based analysis by constructing a 3D surface from stacked 2D slices. This dimensional change enables the system to capture spatial relationships and surface curvature information that are lost in 2D, thereby improving segmentation accuracy and consistency while maintaining processing efficiency through automated 3D surface generation and curvature computation.
2Device complexity
If 2D slice-based curvature computation is used, then computational complexity is reduced, but segmentation consistency across slices deteriorates
Solution Approach 1:
The system computes curvatures on a 3D surface constructed from 2D slices rather than individually on each 2D slice. This approach provides a unified framework that maintains segmentation consistency across the entire volume while managing computational complexity through efficient 3D surface algorithms and optimized curvature calculations.
Solution Approach 2:
The patent merges the curvature computation across multiple 2D slices into a single 3D surface analysis. By combining the surface information from all slices into a unified 3D representation, the system achieves consistent segmentation results throughout the volume while reducing redundant computations that would occur if each slice were processed independently.
3Ease of manufacture
If surface interpolation is used for nodule segmentation, then segmentation can be performed, but pleura voxels are incorrectly included
Solution Approach 1:
The patent applies local quality analysis by computing surface curvatures at each point on the 3D surface and comparing them against threshold criteria. This local evaluation allows the system to distinguish between true nodule surfaces and the pleura based on their distinct curvature characteristics, thereby excluding pleura voxels from the segmentation while maintaining the ability to segment nodules.
Data Source
AI summary
A method of detecting irregularities in a digitized image includes selecting a volume of interest in an image, growing a surface from a point in said volume of interest, determining the local gradient of each point on said surface, and applying a local filter to said gradient, determining a histogram of filtered gradients, and detecting an irregularity by selecting those points corresponding to gradient histogram values above a threshold value.