Neural Network Scatter Correction for Spectral CT Imaging

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current CT imaging technologies face challenges in accurately correcting scatter artifacts, particularly in spectral or dual-energy CT, which are sensitive to noise and energy-dependent scatter, leading to degraded image quality.

Innovation Solution

A neural network-based approach is employed to generate scatter estimates from CT projection data or uncorrected reconstructed images, using Monte Carlo simulation to optimize the network for accurate scatter correction, enabling efficient and precise removal of scatter artifacts in both projection and image spaces.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If Monte Carlo simulation is used for scatter correction, then accuracy of scatter correction is improved, but computational cost increases

Engineering Contradiction:

Improveaccuracy of scatter correctionVSAvoidcomputational cost

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent pre-calculates scatter estimates using Monte Carlo simulation for various phantom configurations and stores them in lookup tables before actual imaging. During clinical scanning, the pre-computed scatter estimates are retrieved and applied directly, avoiding real-time Monte Carlo computation while maintaining high accuracy

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent creates simplified computational models that replicate the scattering behavior observed in Monte Carlo simulations. These models use reduced complexity representations of patient anatomy and scattering physics, enabling fast scatter correction that closely mimics full Monte Carlo results

Inventive Principle:

Principle #26Copying

2Loss of information

If spectral or dual-energy CT is used to extract additional information, then diagnostic information is improved, but sensitivity to scatter artifacts increases

Engineering Contradiction:

Improvediagnostic informationVSAvoidsensitivity to scatter artifacts

Core Design Contradiction:

Loss of informationVSObject-affected harmful factors

Solution Approach 1:

The patent separates scatter correction into energy-specific components by processing low-energy and high-energy projection data independently. Scatter estimates are calculated and applied separately for each energy range before spectral decomposition, preventing scatter from contaminating the spectral information extraction process

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies energy-dependent scatter correction parameters that are specifically tailored for spectral CT. By adjusting scatter estimation parameters according to the energy spectrum being analyzed, the method maintains high accuracy for diagnostic information extraction while compensating for increased scatter sensitivity

Inventive Principle:

Principle #35Parameter changes

3Object-generated harmful factors

If anti-scatter grids are used to reduce scatter, then scatter reduction is improved, but equipment cost and system complexity increase

Engineering Contradiction:

Improvescatter reductionVSAvoidequipment cost and system complexity

Core Design Contradiction:

Object-generated harmful factorsVSDevice complexity

Solution Approach 1:

The patent replaces physical anti-scatter grids with computational scatter correction methods. By using algorithms that model and subtract scatter from projection data, the system achieves scatter reduction without requiring additional mechanical hardware, thereby reducing equipment cost and system complexity

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

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 method provides more accurate and computationally efficient scatter correction, effectively addressing amplified scatter in spectral images and adapting to specific CT scanner configurations and patient characteristics, while maintaining high accuracy comparable to Monte Carlo simulations.

Implementation Method 1

A neural network-based approach is employed to generate scatter estimates from CT projection data or uncorrected reconstructed images

Methodology Applied

Scientific EffectNeural network processing:

Implementation Method 2

enabling efficient and precise removal of scatter artifacts in both projection and image spaces

Methodology Applied

Scientific EffectScatter correction:

Data Source

PatentUS11769277B2Deep learning based scatter correction

Publication Date: 2023.09.26 KONINKLIJKE PHILIPS NV

AI summary

An imaging system includes a computed tomography (CT) imaging device (10) (optionally a spectral CT), an electronic processor (16, 50), and a non-transitory storage medium (18, 52) storing a neural network (40) trained on simulated imaging data (74) generated by Monte Carlo simulation (60) including simulation of at least one scattering mechanism (66) to convert CT imaging data to a scatter estimate in projection space or to convert an uncorrected reconstructed CT image to a scatter estimate in image space. The storage medium further stores instructions readable and executable by the electronic processor to reconstruct CT imaging data (12, 14) acquired by the CT imaging device to generate a scatter-corrected reconstructed CT image (42). This includes generating a scatter estimate (92, 112, 132, 162, 182) by applying the neural network to the acquired CT imaging data or to an uncorrected CT image (178) reconstructed from the acquired CT imaging data.