X-ray Noise Estimation via Alternating Negation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for estimating image noise in volumetric image data, such as in CT scans, reduce angular or radial resolution and result in inaccurate noise estimates due to the splitting of views into odd and even sets, leading to suboptimal de-noising.

Innovation Solution

An alternating negation approach is used to process projection data, where views are multiplied by alternating positive and negative factors to generate a noise-only image, allowing for accurate noise estimation without reducing resolution.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If projection data is split into odd and even views for noise estimation, then noise image generation is enabled, but angular or radial resolution is reduced

Engineering Contradiction:
Improvenoise estimation accuracyVSAvoidangular or radial resolution
Core Design Contradiction:
Measurement precisionVSShape

Solution Approach 1:

Instead of splitting views into odd and even sets to estimate noise, the patent applies alternating negation by multiplying projection data by alternating positive and negative factors (e.g., +1, -1, +1, -1). This inversion approach allows the entire projection data to be used for noise estimation while maintaining full angular and radial resolution, as no data is discarded or subsampled.

Inventive Principle:
Principle #13The other way round (Inversion)

Solution Approach 2:

The patent changes the parameter approach from view-splitting (odd/even separation) to alternating sign multiplication. By applying alternating positive and negative factors to the projection data, the method transforms the noise estimation process into a full-resolution operation, eliminating the resolution loss inherent in view-splitting methods.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If views are split into odd and even sets for noise estimation, then noise image can be generated, but noise estimates become inaccurate

Engineering Contradiction:
Improvenoise estimation accuracyVSAvoidnoise estimate reliability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent inverts the conventional approach by using alternating negation on the full projection data set. This allows accurate noise estimation by preserving all view information and avoiding the inaccuracies introduced by odd/even view splitting, where structural information is lost and noise estimates become unreliable.

Inventive Principle:
Principle #13The other way round (Inversion)

3Measurement precision

If projection data is processed through multiple processing chains for noise estimation, then noise image is generated, but device complexity increases

Engineering Contradiction:
Improvenoise estimation accuracyVSAvoidprocessing chain complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent merges the noise estimation process into a single processing chain that operates on the full projection data using alternating negation. Instead of requiring separate odd and even view processing chains, the method combines all views and applies alternating positive and negative factors, simplifying the overall system architecture while maintaining noise estimation accuracy.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP3552181B1Image noise estimation using alternating negation
Publication Date: 2021.11.10 KONINKLIJKE PHILIPS NV
  • EP3552181B1 patent drawingFigure 1~2
  • EP3552181B1 patent drawingFigure 3
  • EP3552181B1 patent drawingFigure 4~5

AI summary

An imaging system (400) includes a radiation source (408) configured to emit X-ray radiation, a detector array (410) configured to detected X-ray radiation and generate projection data indicative thereof, and a first processing chain (418) configured to reconstruct the projection data and generate a noise only image. A method includes receiving projection data produced by an imaging system and processing the projection data with a first processing chain configured to reconstruct the projection data and generate a noise only image. A processor is configured to: scan an object or subject with an x-ray imaging system and generating projection data, process the projection data with a first processing chain configured to reconstruct the projection data and generate a noise only image, process the projection data with a second processing chain configured to reconstruct the projection data and generate a structure image, and de-noise the structure image based on the noise only image.