Dual-Energy Radiation Image Processing for Position-Corrected Bone Density
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods struggle to accurately derive composition information of a subject due to changes in positioning and deviation of radiation incidence angles, leading to inaccurate comparisons between radiation images captured at different positions.
Innovation Solution
An image processing device and learning device that utilize a trained model to derive a deviation angle and composition information by processing first and second radiation images, removing scattered ray components, and performing weighting subtraction to extract bone part images, using a neural network trained on three-dimensional image data to correct for positional deviations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the subject positioning is performed accurately at the time of imaging, then the composition information can be derived accurately, but it takes time to perform the imaging
Solution Approach 1:
The system performs preliminary actions by acquiring multiple radiation images at different positions before final analysis. A neural network is trained in advance to recognize deviation angles and correct composition information, so that when actual imaging occurs, the system can quickly process pre-acquired data without requiring time-consuming real-time positioning adjustments
Solution Approach 2:
The system implements feedback mechanisms where the neural network continuously learns from radiation images to correct deviation angles and composition information. The feedback loop allows the system to automatically adjust and refine measurements based on actual imaging data, improving accuracy without requiring additional manual positioning time
2Measurement precision
If the radiation is emitted from the front of the subject, then the composition information can be derived, but the incidence angle deviates when the subject rotates, leading to inaccurate comparison
Solution Approach 1:
The system changes parameters by dynamically calculating deviation angles based on the actual position of the subject in each radiation image. Instead of assuming a fixed front-incident angle, the neural network computes the actual angle and uses this corrected parameter to derive accurate composition information, making the system robust to various positioning variations
Solution Approach 2:
The system creates a virtual copy of the subject's anatomical structure through the neural network model. This digital replica allows the system to simulate and correct for angle deviations by comparing the actual radiation image with the modeled structure, enabling accurate composition derivation regardless of the subject's actual positioning
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
Enables accurate derivation of composition information, such as bone density, regardless of subject positioning, by correcting for radiation angle deviations and removing scattered rays, thereby improving image analysis precision.
Implementation Method 1
remove scattered ray components of the first radiation image and the second radiation image to derive a first primary ray image and a second primary ray image
Implementation Method 2
energy subtraction processing using two radiation images obtained by irradiating a subject with two types of radiation having different energy distributions by using the fact that an attenuation amount of the transmitted radiation differs depending on the substance constituting the subject
Data Source
AI summary
A processor acquires a structure image representing at least one structure in a subject based on at least one radiation image of the subject, and derives a deviation angle of the structure included in the structure image with respect to a reference position and composition information of the structure at the reference position by using a trained model that outputs an estimation result of a deviation angle of radiation with respect to the reference position for the structure included in the structure image and the composition information of the structure at the reference position by input of the structure image.