Iterated Coordinate Descent for EDCT Material Decomposition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current CT systems face challenges in effectively handling material decomposition for energy discriminating computed tomography acquisitions, particularly in reducing noise and improving separation of material components using traditional reconstruction methods.
Innovation Solution
The method extends iterated coordinate descent (ICD) optimization to handle material decomposition for energy discriminating computed tomography (EDCT) by obtaining current path length estimates, performing a sequence of iterations, and solving small dimensional systems of linear equations to update pixel component values, incorporating prior information and achieving convergence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional reconstruction algorithms are used for energy discriminating computed tomography, then the processing speed is maintained, but the noise reduction and material component separation are insufficient
Solution Approach 1:
The patent segments the material decomposition problem into multiple iterations where each iteration updates one material component at a time while holding others fixed. This coordinate descent approach divides the complex multi-material decomposition into simpler single-material optimization steps, improving both separation quality and maintaining computational efficiency
Solution Approach 2:
The patent implements an iterative dynamic reconstruction process where the solution is progressively refined through multiple passes. Each iteration dynamically adjusts the material component estimates based on current projections, allowing the system to converge to a better solution without requiring static batch processing
2Measurement precision
If iterative reconstruction techniques are applied to reduce image noise, then the measurement precision improves, but the processing time increases
Solution Approach 1:
The patent applies partial iteration by performing a fixed number of iterations or stopping when convergence criteria are met, rather than completing exhaustive iterations. This allows the system to achieve sufficient noise reduction without the full computational cost of complete convergence, balancing image quality with processing time
Solution Approach 2:
The patent uses preliminary actions such as analytical pre-processing steps and initial estimate generation before applying iterative refinement. This preliminary preparation reduces the burden on subsequent iterative steps, allowing faster convergence with fewer iterations and thus reducing overall processing time while maintaining noise reduction benefits
3Reliability
If direct material decomposition techniques are used, then the processing is faster, but the separation of material components and robustness to data quality problems deteriorates
Solution Approach 1:
The patent implements feedback loops where each iteration uses the current material component estimates to generate synthetic projections, compares these with actual measured projections, and uses the difference (residual) to update the estimates. This feedback mechanism continuously refines the solution and makes it robust to data quality issues by progressively correcting errors
Solution Approach 2:
The iterative algorithm is self-correcting, automatically adjusting material component estimates based on inconsistencies in the data. The system serves itself by identifying and correcting its own errors through the optimization process, improving robustness without external intervention while maintaining reasonable processing speed through efficient update rules
Data Source
AI summary
A new method extends iterated coordinate descent (“ICD”)—an optimization method employed in some statistical reconstruction algorithms—to handle material decomposition (“MD”) for energy discriminating computed tomography (“EDCT”) acquisitions.