Neural Network Estimation of Composition-Specific Images from Multi-Energy Radiation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating images that emphasize specific compositions, such as bone parts, from radiation images lack accuracy due to limitations in processing techniques and data utilization.
Innovation Solution
An estimation device employing a learned neural network that uses two radiation images with different energy distributions and derived emphasis images to accurately estimate specific compositions, such as bone or soft parts, through energy subtraction processing and scattered ray removal techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If energy subtraction processing is used to emphasize specific compositions, then the ability to derive composition information is improved, but the accuracy of estimation is insufficient
Solution Approach 1:
The patent introduces a new dimension by using multiple radiation images with different energy distributions (e.g., low-energy and high-energy images) instead of relying on a single image. This multi-energy approach adds dimensional information that enables more accurate derivation of composition-specific images through energy subtraction processing, thereby improving both information extraction and estimation accuracy simultaneously.
Solution Approach 2:
The patent changes the energy distribution parameter of the radiation by acquiring images at multiple energy levels. This parameter change allows the system to exploit the different attenuation characteristics of various tissues at different energies, enabling more precise separation and estimation of specific compositions like bone and soft tissue while maintaining high accuracy.
2Ease of operation
If simple imaging is used to acquire radiation images, then the imaging process is simplified, but the accuracy of deriving specific composition images is insufficient
Solution Approach 1:
The patent performs preliminary action by acquiring multiple radiation images with different energy distributions before deriving the specific composition images. This pre-acquisition of multi-energy data provides the necessary foundation for accurate energy subtraction processing, enabling high-accuracy estimation without complicating the actual imaging operation during the examination process.
Solution Approach 2:
The patent creates copies of the subject from different energy perspectives by acquiring multiple radiation images at different energy levels. These copies serve as the basis for energy subtraction processing, allowing the system to reconstruct specific composition images with high accuracy while maintaining operational simplicity through automated processing of these energy-based copies.
3Extent of automation
If learned models are used to derive bone part images, then the processing is automated, but the accuracy is limited compared to energy subtraction methods
Solution Approach 1:
The patent introduces energy subtraction processing as an intermediary step between acquiring raw radiation images and deriving final composition-specific images. This intermediary process uses the physical principle of differential attenuation at different energies to separate bone and soft tissue information, providing a more accurate intermediate representation than direct learned model approaches while maintaining automation through computational processing.
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
The solution enables high-accuracy estimation of images emphasizing specific compositions, improving upon existing methods by utilizing advanced neural network processing and energy subtraction techniques.
Implementation Method 1
energy subtraction processing of performing weighting subtraction on the two radiation images
Implementation Method 2
deriving attenuation coefficients of the bone part and the soft part by using results of recognition of the bone part and the soft part and the two radiation images
Data Source
AI summary
An estimation device includes at least one processor, in which the processor functions as a learned neural network that derives a result of estimation of at least one emphasis image in which a specific composition of a subject including a plurality of compositions is emphasized from a simple two-dimensional image acquired by simply imaging the subject. The learned neural network is learned by using, as teacher data, two radiation images acquired by imaging the subject with radiation having different energy distributions and an emphasis image for learning in which the specific composition of the subject is emphasized, which is derived from the two radiation images.