X-ray Unit Pose Determination Using Digital Model and Attenuation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current X-ray devices face challenges in achieving accurate 3D tomographic reconstruction due to inaccuracies in the configuration of the X-ray device, particularly in mobile systems where the positions of the X-ray source, detector, and object are arbitrary, leading to poor image quality.
Innovation Solution
A method that determines the pose of the X-ray unit relative to the object using a digital model and machine learning, specifically an artificial neural network, to improve tomographic reconstruction by calculating model values for various poses and training the network to recognize relationships between the X-ray unit's pose and attenuation, allowing for precise determination of the X-ray unit's position and orientation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If mobile X-ray systems are used with arbitrary positions of X-ray source, detector, and object, then ease of operation and adaptability are improved, but measurement precision and manufacturing precision of the configuration deteriorate
Solution Approach 1:
The patent replaces mechanical calibration systems with a computational approach using deep learning neural networks. The system substitutes physical measurement and alignment mechanisms with digital image processing and machine learning algorithms that automatically determine geometric parameters from X-ray images, eliminating the need for complex mechanical calibration while maintaining precision.
Solution Approach 2:
The patent uses calibration objects with known geometric structures (such as spheres, cylinders, or specific patterns) that create distinctive, easily identifiable images. These calibration objects serve as reference copies with predetermined properties that allow the system to calculate accurate geometric parameters through image analysis without requiring precise mechanical positioning.
2Measurement precision
If calibration blocks and self-calibration methods are used, then measurement precision is improved, but device complexity and loss of time are worsened
Solution Approach 1:
The patent implements self-calibration through automated deep learning algorithms that perform calibration without external intervention. The neural network automatically analyzes calibration object images, determines geometric parameters, and configures the reconstruction algorithm, eliminating the need for manual calibration procedures and reducing dependency on complex external calibration equipment.
Solution Approach 2:
The patent changes the approach from fixed mechanical calibration parameters to dynamic computational parameters. The system uses variable geometric parameters derived from image analysis rather than fixed mechanical settings, allowing flexible adaptation to different configurations while simplifying the overall system architecture.
3Measurement precision
If calibration blocks and self-calibration methods are used, then measurement precision is improved, but loss of time is worsened
Solution Approach 1:
The patent performs preliminary calibration by pre-training deep learning models with extensive calibration data before actual measurements. The system prepares geometric parameter calculation algorithms in advance, so that during actual operation, calibration can be performed rapidly using pre-processed models rather than requiring time-consuming real-time calculations or manual procedures.
Solution Approach 2:
The patent replaces time-consuming manual calibration and iterative optimization procedures with automated deep learning inference. The neural network rapidly determines geometric parameters from calibration images in seconds, eliminating the time required for manual measurement, calculation, and verification steps associated with traditional calibration methods.
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
This approach enables improved tomographic reconstruction by accurately determining the pose of the X-ray unit relative to the object, enhancing the quality of reconstructed images and minimizing errors, even in non-fixed configurations.
Implementation Method 1
an X-ray beam is first emitted by an X-ray unit (1) comprising an X-ray detector (3) and an X-ray source (2) through which the X-ray beam (5) is emitted
Implementation Method 2
the attenuation of the X-ray beam (5) during transmission of the object (4) located in a beam path of the X-ray beam (5) of the X-ray unit (1) is determined
Data Source
Figure 1~2
AI summary
The invention relates to a method for enabling a tomographic reconstruction, comprising the following steps: - emitting an x-ray beam (5) through an x-ray unit (1), - determining a damping of the x-ray beam (5) during transmission of the object (4) located in a beam path of the x-ray beam (5) of the x-ray unit (1), - determining structure data of the object (4) using the damping of the x-ray beam (5). To enable the tomographic reconstruction of the object to be improved, the invention proposes that - using a digital model of the object (4) and using the damping of the x-ray beam (5), a pose of the x-ray unit (1) relative to the object (4) is determined to facilitate the tomographic reconstruction.