Multi-Energy CT Material Identification via Segmented Decomposition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multi-energy CT technologies face challenges in accurately identifying and quantifying a large number of different materials due to numerical instability and material misidentification, particularly when dealing with similar x-ray attenuation properties and poor data quality.
Innovation Solution
A computer program product and system that employs a method for selecting appropriate material decomposition algorithms based on voxel classification into air, low density, or high density categories, using metrics like Mahalanobis distance and Euclidean distance, and applies specific decomposition methods to each category, along with sparse solution enforcement and rejection criteria to identify materials effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a large number of materials are included in the material dictionary for decomposition, then material identification completeness is improved, but numerical stability deteriorates due to ill-conditioned inversion
Solution Approach 1:
The patent segments the material decomposition problem by classifying voxels into different density categories (air, low density, high density) and applying different decomposition strategies to each category. This segmentation allows the system to handle a large material dictionary without suffering from numerical instability across all voxels, as each category can be processed with appropriate constraints and methods.
Solution Approach 2:
The patent applies different decomposition methods and constraints to different voxel types based on their local properties. Low density voxels use one set of constraints while high density voxels use another set, allowing each local region to be processed with the most appropriate method for its characteristics, thereby maintaining numerical stability while preserving material identification accuracy.
2Productivity
If material decomposition is performed on poor quality data with significant image artefacts, then processing speed is maintained, but measurement precision deteriorates due to excessive noise
Solution Approach 1:
The patent changes processing parameters based on voxel characteristics, applying different decomposition methods and constraint strengths to different density categories. This allows the system to maintain processing speed by using efficient methods for simple cases while applying more robust, noise-resistant methods to challenging voxels with artefacts, thereby preserving measurement precision without sacrificing productivity.
3Reliability
If materials are omitted from the decomposition to achieve stable inversion, then numerical stability is improved, but material identification accuracy deteriorates due to projection onto given basis
Solution Approach 1:
The patent makes the material decomposition dynamic by allowing different numbers and types of materials to be decomposed in different voxel regions based on their density characteristics. Rather than using a fixed material basis for all voxels, the system adaptively selects which materials to decompose based on local voxel properties, maintaining inversion stability while preserving the ability to accurately identify all materials present.
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 the accurate identification and quantification of multiple materials within multi-energy CT scans, improving stability and accuracy by segregating material decomposition problems and enforcing sparse solutions, thus overcoming limitations of existing methods.
Implementation Method 1
x-rays are absorbed differently by different materials in the body, and differently again for x-rays of different energies. Multi-energy CT measures the absorption of x-rays in different energy ranges.
Data Source
Figure 1(a)~1(b)
Figure 2
Figure 3
AI summary
Disclosed are methods for identification and quantification of a number of different materials within an object using one or more multi-energy CT imaging devices and the image data sets produced therefrom. Identification and quantification of different materials is achieved by using the following three properties: solve only for sparse solutions; separate the soft tissue problem from the dense material problem; and use a combinatorial approach to allow for simple application of different constraints to different combinations of materials. Also disclosed are one or more computer program products, computer systems or computer implemented methods for the identification of multiple materials within an object.