Anti-correlation Filter for Spectral CT Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In spectral computed tomography, anti-correlated noise in projection data leads to streak artifacts and reduced clinical value of reconstructed images due to noise amplification, and existing anti-correlation filters can cause crosstalk between basis material data sets, reducing diagnostic value.
Innovation Solution
An anti-correlation filter with regularization terms and corresponding scaling factors is applied to basis material line integrals, balancing the regularization effect across different materials to mitigate crosstalk and noise amplification, using a regularized maximum likelihood algorithm to produce de-noised basis material line integrals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If an anti-correlation filter is applied to basis material line integrals to reduce anti-correlated noise, then noise reduction is improved, but crosstalk between basis material data sets occurs causing artifacts
Solution Approach 1:
The patent modifies the regularization parameters by introducing material-specific scaling factors that adjust the strength of regularization applied to each basis material. This parameter change allows differential treatment of different materials during filtering, reducing crosstalk artifacts while maintaining noise reduction benefits. The scaling factors are determined based on material properties and noise characteristics, enabling optimized filtering for each material type.
Solution Approach 2:
The patent applies local quality by making the regularization strength material-dependent rather than uniform across all basis materials. Each basis material receives a customized regularization strength through its specific scaling factor, allowing the filter to adapt locally to the characteristics of each material. This localized approach prevents over-regularization of certain materials that causes crosstalk while maintaining effective noise suppression.
2Reliability
If regularization is applied to control noise in basis material line integrals, then noise amplification is reduced, but crosstalk between different materials increases
Solution Approach 1:
The patent introduces material-specific scaling factors as parameter changes that modulate the regularization strength for each basis material independently. By changing the regularization parameters from a uniform value to material-dependent values, the system achieves better noise control for each material without inducing crosstalk. The scaling factors are optimized based on the noise characteristics and physical properties of each basis material.
3Measurement precision
If strong regularization is applied to reduce noise in reconstructed images, then image quality is improved, but diagnostic value decreases due to crosstalk artifacts
Solution Approach 1:
The patent applies local quality by making regularization strength material-dependent through scaling factors. Different basis materials receive different levels of regularization tailored to their specific noise characteristics and diagnostic importance. This localized regularization approach maintains high image quality while preserving diagnostic accuracy by preventing crosstalk artifacts that would otherwise compromise diagnostic value.
Solution Approach 2:
The patent modifies the regularization parameters by introducing material-specific scaling factors that adjust the filtering strength for each basis material. This parameter change enables optimized noise reduction for each material type, improving overall image quality while maintaining diagnostic reliability by avoiding excessive regularization that would create crosstalk artifacts.
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 effectively reduces anti-correlated noise and minimizes crosstalk between tissue boundaries, enhancing the diagnostic value of reconstructed images by balancing regularization terms with scaling factors, resulting in clearer and more accurate volumetric image data.
Implementation Method 1
filtering the at least two sets of noisy basis material line integrals with an anti-correlation filter that at least includes a regularization term with regularization factors
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A method includes receiving at least two sets of noisy basis material line integrals, each set corresponding to a different basis material and filtering the at least two sets of noisy basis material line integrals with an anti-correlation filter that at least includes a regularization term with balancing regularization factors, thereby producing de-noised basis material line integrals. An imaging system (100) includes a projection data processor (116) with an anti-correlation filter (118) that filters at least two sets of noisy basis material line integrals, each set corresponding to a different basis material, thereby producing de-noised basis material line integrals, wherein the anti-correlation filter includes a regularization term with regularization balancing factors.