3D Image Generation via Partial Volume Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital breast tomosynthesis methods require diagnosticians to mentally correlate multiple slice images for 3D tissue diagnosis, leading to challenges in identifying twisted tissues due to depth blurring and overlapping issues in projection images.
Innovation Solution
A method and apparatus that generate and divide a 3D volume into partial volumes using X-ray projection images acquired at varying angles, creating stereoscopic images with adjusted baseline and thickness to reduce depth blurring and overlapping, while optimizing radiation dose and improving signal-to-noise ratio through forward projection and interpolation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If multiple slice images are used for 3D tissue diagnosis, then diagnostic information completeness is improved, but depth blurring and tissue overlap occur making twisted tissues unidentifiable
Solution Approach 1:
The patent divides the full 3D volume into multiple partial volumes (e.g., upper, middle, lower portions) and generates separate projection images for each partial volume. This segmentation allows diagnosticians to view specific regions without depth blurring from other regions, while maintaining complete diagnostic information across all partial volume images.
Solution Approach 2:
The patent transitions from 2D slice images to 3D projection images by synthesizing multiple X-ray projection images taken at different angles. This dimensional transformation provides true 3D visualization where twisted tissues can be identified, eliminating the depth blurring problem inherent in 2D slice representations.
2Measurement precision
If X-ray projection images are acquired at multiple angles for 3D reconstruction, then 3D visualization quality is improved, but radiation dose to the object increases
Solution Approach 1:
The patent acquires X-ray projection images at a limited angular range (e.g., -30° to +30°) rather than requiring a full 180° or 360° rotation. This partial action approach provides sufficient 3D visualization quality for diagnostic purposes while significantly reducing the cumulative radiation dose compared to conventional tomosynthesis or CT scanning.
3Area of stationary object
If the full 3D volume is visualized, then complete tissue coverage is achieved, but apparent overlap and depth blurring increase reducing diagnostic clarity
Solution Approach 1:
The patent segments the full 3D volume into multiple partial volumes and generates separate projection images for each segment. This allows diagnosticians to examine specific regions of interest without the obscuring overlap and depth blurring that occur when the entire volume is displayed simultaneously, thereby improving diagnostic clarity while maintaining complete tissue coverage across all segments.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enhances diagnostic accuracy by providing clear, 3D visualizations with reduced depth blurring and overlapping, allowing for better tissue identification and improved diagnostic environments with enhanced signal-to-noise ratio.
Implementation Method 1
creating a plurality of first projection images by detecting, by using an X-ray detector, X-rays which are emitted from an X-ray source toward an object at an angle from within a predetermined angular range
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided is a method and apparatus for providing a three-dimensional (3D) image. A plurality of first projection images may be created by detecting X-rays which are emitted toward an object at different angles. A plurality of second projection images with respect to a partial volume of the object may be created by applying a forward projection and interpolation to at least one of the plurality of first projection images. A left image and a right image may be selected from among the plurality of second projection images, and the selected left image and right image may be displayed for a user as a 3D projection image.