Iterative CT Reconstruction for Truncation Artifacts

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

CT imaging systems face challenges in accurately reconstructing images due to data truncation, which leads to inaccurate projection measurements and image artifacts, especially when objects extend beyond the scan field-of-view, causing fluctuations in x-ray tube current and angular dependence of detector gain.

Innovation Solution

The method involves iterative reconstruction algorithms that jointly estimate error correction gain parameters and the reconstructed image, using a weighted least-squares approach to remove low-frequency shading artifacts by normalizing sinogram data with the X-ray tube generator signal and performing DC component removal within the iterative loop.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If pre-processing corrections are applied to normalize projections using reference channels, then image artifacts from x-ray tube current fluctuations are reduced, but when objects extend beyond the scan field-of-view, the reference channels are blocked and correction coefficients become inaccurate, resulting in view-dependent DC bias and image artifacts

Engineering Contradiction:
Improveaccuracy of correction coefficientsVSAvoidimage artifacts from data truncation
Core Design Contradiction:
ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent applies preliminary DC bias correction to the projection data before iterative reconstruction. By estimating and removing the view-dependent DC bias component in advance, the method prepares the data to be less sensitive to truncation errors, allowing iterative reconstruction to converge to an accurate solution even when reference channels are blocked

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent incorporates a feedback mechanism where the iterative reconstruction process continuously refines the estimate of the object and the DC bias components. The reconstruction algorithm uses the current estimate to update the DC bias correction, which in turn improves the next iteration's reconstruction, creating a self-correcting system that handles truncated data

Inventive Principle:
Principle #23Feedback

2Measurement precision

If iterative reconstruction algorithms are used to improve image quality over conventional FBP, then image quality is significantly improved, but using the same pre-processed data to initialize iterative reconstruction results in significant image artifacts in cases where projection data is truncated

Engineering Contradiction:
Improveimage qualityVSAvoidaccuracy of reconstructed image
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent segments the reconstruction problem into two independent components: the object being imaged and the DC bias component. By separating these components and reconstructing them independently with appropriate regularization, the method avoids the artifacts that occur when truncated data is processed as a single unified reconstruction problem

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different regularization strategies to different components of the reconstruction. The object reconstruction uses standard regularization appropriate for anatomical structures, while the DC bias component uses regularization tailored to its smooth, low-frequency nature. This localized approach to quality control allows each component to be reconstructed with optimal accuracy

Inventive Principle:
Principle #3Local quality

3Measurement precision

If reference channels are placed outside the scan field-of-view to measure x-ray photons directly from the x-ray tube, then x-ray flux can be monitored for normalization, but when objects are present outside the scan field-of-view, the reference channels are blocked and accurate estimation of correction coefficients becomes impossible

Engineering Contradiction:
Improvex-ray flux measurement accuracyVSAvoidrobustness to objects outside field-of-view
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent introduces an intermediary mathematical model that relates the measured projection data to the underlying x-ray flux and object attenuation. Instead of directly measuring flux with vulnerable reference channels, the method uses the projection measurements themselves as intermediaries, combined with a physical model of x-ray attenuation, to infer the flux variations and correct for them

Inventive Principle:
Principle #24Intermediary (Mediator)

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 significantly reduces low-frequency shading artifacts in iterative reconstruction images, improving image quality and accuracy by accounting for truncation-related errors and maintaining stable low-frequency content throughout the reconstruction process.

Implementation Method 1

a radiation source projects a fan-shaped beam which is collimated to lie within an X-Y plane

Methodology Applied
Scientific EffectX-ray generation: X-Ray

Implementation Method 2

The intensity of the attenuated radiation beam received at the detector array is dependent upon the attenuation of a radiation beam by the object

Methodology Applied
Scientific EffectX-ray attenuation: Absorption (EM radiation)

Implementation Method 3

Each detector element of the array produces a separate electrical signal that is a measurement of the beam attenuation at the detector location

Methodology Applied
Scientific EffectRadiation detection: Photoelectric Effect

Data Source

PatentEP2149284B1Methods and systems to facilitate correcting gain fluctuations in image
Publication Date: 2014.08.13 GENERAL ELECTRIC CO
  • EP2149284B1 patent drawingFigure 1~2
  • EP2149284B1 patent drawingFigure 3A~3B
  • EP2149284B1 patent drawingFigure 4A~4B

AI summary

Methods and systems for reconstructing an image are provided. The method includes performing a tomographic image reconstruction using a joint estimation of at least one of a gain parameter and an offset parameter, and an estimation of the reconstructed image.