Biomedical Image Reconstruction Using Correction Matrices

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for reconstructing biomedical images using compressed sensing are limited in utilizing redundant information for artifact correction and do not effectively incorporate prior knowledge of artifacts, especially when multiple sets of image data are acquired under different undersampling schemes or modes, leading to incomplete artifact suppression.

Innovation Solution

The method involves acquiring two or more sets of image data using different undersampling schemes or acquisition modes, multiplying each data set by a correction matrix calculated from a mathematical model of expected artifacts, and using a nonlinear iterative algorithm to combine the data sets, ensuring fidelity to corrected acquired data and suppressing incoherent artifacts.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If multiple sets of image data are acquired using different undersampling schemes to suppress artifacts, then image quality is improved, but the complexity of the reconstruction algorithm increases

Engineering Contradiction:
Improveimage qualityVSAvoidreconstruction algorithm complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The reconstruction algorithm is segmented into distinct functional modules: a combination module that merges multiple undersampled data sets, a artifact suppression module that identifies and removes incoherent artifacts, and a reconstruction module that generates the final image. This modular segmentation makes the complex algorithm more manageable and implementable while achieving superior image quality through systematic artifact suppression.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If multiple sets of image data are acquired with different undersampling schemes, then artifact suppression is improved, but acquisition time increases

Engineering Contradiction:
Improveartifact suppressionVSAvoidacquisition time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

Multiple data sets are acquired by changing the undersampling parameters (different sampling patterns, different acceleration factors) rather than acquiring completely separate data sets. This allows the system to exploit parameter variability to make artifacts incoherent across data sets, improving artifact suppression while minimizing the increase in acquisition time by reusing the same basic acquisition protocol with modified parameters.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If a correction matrix is applied to each data set based on mathematical models of expected artifacts, then fidelity to corrected data is improved, but computational complexity increases

Engineering Contradiction:
Improvefidelity to corrected dataVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

Correction matrices are pre-calculated based on mathematical models of expected artifacts before the actual reconstruction process. These pre-computed correction matrices account for common artifact patterns, allowing the system to quickly apply corrections during reconstruction without performing complex real-time calculations. This preliminary action maintains high fidelity to corrected data while reducing the computational burden during the actual reconstruction phase.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2660618B1Biomedical image reconstruction method
Publication Date: 2020.10.14 ESAOTE
  • EP2660618B1 patent drawingFigure 1
  • EP2660618B1 patent drawingFigure 2a~2b
  • EP2660618B1 patent drawingFigure 3a~3c

AI summary

A method of reconstructing an MRI or ultrasound biomedical image, based on compressed sensing and comprising the steps of acquiring several sets of image data from said generated signals, each data set being acquired in a different undersampling scheme and/or a different acquisition mode such as to make expected and unavoidable artifacts incoherent. Each of the acquired image data set is multiplied by a correction matrix Δ, which is calculated from a mathematical model of expected artifacts according to prior knowledge, for adjusting fidelity of the reconstructed image to the corrected acquired image data. For each iteration of said nonlinear iterative algorithm the data sets are processed to generate a combination image which will therefore be faithful to the acquired data but not to the incoherent artifacts.