CT Image Reconstruction Overlay for Noise and Visual Clue Balance

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

De-noising algorithms in reconstructed image data, such as those used in computed tomography, remove visual clues that radiologists rely on for confidence and resolution assessment, leading to reluctance in using these algorithms.

Innovation Solution

Concurrently present reconstructed image data with varying noise levels to maintain visual clues, allowing for better feature identification by overlaying or combining images processed with different reconstruction algorithms.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If de-noising reconstruction algorithms are used, then noise in reconstructed image data is reduced, but visual clues that radiologists rely on for confidence and resolution assessment are removed

Engineering Contradiction:
Improvenoise reductionVSAvoidloss of visual clues
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent divides the reconstructed image data into multiple representations: the original reconstructed image data and one or more de-noised versions. Each representation preserves different characteristics - the original retains visual clues while the de-noised version reduces noise. This segmentation allows radiologists to access both types of information separately or in combination, resolving the contradiction between noise reduction and preservation of visual assessment clues.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If de-noising algorithms are applied, then image quality is improved by reducing noise, but image characteristics change in ways that radiologists are not trained to interpret

Engineering Contradiction:
Improveimage qualityVSAvoidcompatibility with radiologist training
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent implements a dynamic display system that allows radiologists to switch between different image representations (original and de-noised) and to adjust the degree of de-noising applied. This dynamic adaptability enables radiologists to work with image characteristics that match their training while still benefiting from noise reduction capabilities, resolving the contradiction between improved image quality and compatibility with existing expertise.

Inventive Principle:
Principle #15Dynamics

3Measurement precision

If noise is reduced in reconstructed image data, then features obscured by noise become more visible, but visual cues for tissue discrimination are removed

Engineering Contradiction:
Improvefeature visibilityVSAvoidloss of tissue discrimination cues
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent introduces an intermediary display mechanism that overlays or juxtaposes de-noised image regions with original image regions. This intermediary presentation allows radiologists to simultaneously observe both the enhanced feature visibility from de-noising and the tissue discrimination cues from the original noisy image, effectively mediating between the two competing needs and resolving the contradiction.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP2984630B1Reconstructed image data visualization
Publication Date: 2026.05.06 KONINKLIJKE PHILIPS NV
  • EP2984630B1 patent drawingFigure 1
  • EP2984630B1 patent drawingFigure 2
  • EP2984630B1 patent drawingFigure 3

AI summary

A method includes processing projection data with a first reconstruction algorithm and reconstructing first reconstructed volumetric image data, wherein the first reconstructed volumetric image data has a first 3D noise function. The method further includes processing the same projection data with a second different reconstruction algorithm and reconstructing second reconstructed volumetric image data, wherein the second reconstructed volumetric image data has a second 3D noise function, which is different from the first3Dnoise function. The method further includes visually presenting the first or the second reconstructed volumetric image data in a main viewport. The method further includes visually presenting a sub-portion the other of the first or the second reconstructed volumetric image data in a region of interest overlaid over a sub-portion of the main viewport.