MPI Image Reconstruction Using Eigen-Subtraction for Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for reducing ghosting and blurring artifacts in multi-color magnetic particle imaging (MPI) reconstruction are computationally expensive and often ineffective, leading to misclassification of particle signals and poor tracer quantification.
Innovation Solution
A method utilizing an appended system matrix and eigen-reconstruction of selected voxels to subtract artifacts, combined with optional weighting to match intensity, iteratively refining the image data to reduce ghosting and blurring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multi-color reconstruction is used to separate different particle classes, then the ability to distinguish between particle classes is improved, but ghosting artifacts arise when tracers are not distinctive enough
Solution Approach 1:
The patent extracts and removes ghosting artifacts from the reconstructed image data by identifying and subtracting the artifact components from the total signal. This is achieved by separating the true signal from the ghosting artifacts through mathematical operations on the system matrix and measured signal, thereby eliminating the harmful factor while preserving the useful information about different particle classes.
Solution Approach 2:
The patent converts the harmful ghosting artifacts into beneficial information by using the artifact patterns to identify and characterize the particle classes. The ghosting artifacts, which initially represent misclassification errors, are transformed into useful signals that help distinguish between different particle types through their characteristic patterns in the reconstructed images.
2Measurement precision
If system-function-based MPI image reconstruction is used, then the relationship between measurement signal and particle distribution is improved, but blurring artifacts arise in the reconstructed images
Solution Approach 1:
The patent extracts and removes blurring artifacts from the reconstructed image data by identifying and subtracting the artifact components from the total signal. This is achieved by separating the true signal from the blurring artifacts through mathematical operations on the system matrix and measured signal, thereby eliminating the harmful factor while preserving the useful information about particle distribution.
Solution Approach 2:
The patent applies inversion operations to reverse the blurring effect caused by the system function. By inverting the system matrix or using iterative reconstruction methods, the patent recovers the sharp image details that were lost during the initial reconstruction, thereby converting the blurred image into a clear representation of particle distribution.
3Adaptability or versatility
If multiple system matrices are acquired for different particle classes, then the ability to reconstruct separate images for each tracer type is improved, but computational complexity increases
Solution Approach 1:
The patent merges multiple system matrices into a single appended system matrix that contains information about all particle classes. This consolidation reduces the computational complexity by eliminating the need to manage and process multiple separate system matrices, while still maintaining the ability to reconstruct separate images for each tracer type through the integrated matrix structure.
Solution Approach 2:
The appended system matrix serves multiple functions simultaneously: it acts as the system matrix for all particle classes, provides the basis for artifact removal operations, and enables multi-color reconstruction. This multi-functionality reduces the overall system complexity by replacing multiple specialized matrices with a single universal matrix that handles all reconstruction tasks.
Data Source
AI summary
A method for assessing and/or reducing artifacts arising from reconstruction of magnetic particle imaging (MPI) total image data set (c(r)) from MPI signal data (s(f), u(t)) using an appended system matrix (SM_AB) comprises:a) reconstructing signal data (s(f), u(t)) using appended system matrix (SM_AB) derived from at least two system matrices (SM_A, SM_B) for generating input image data (cin(r));b) selecting at least one voxel as selected voxel from the input image data (cin(r)) for generating output image data (cout(r));c) determining eigen-reconstruction signal data sen(f) by selecting one or more line entries of the appended system matrix (SM_AB) that correspond to the selected voxel(s);d) reconstructing eigen-reconstruction signal data se(f) using the appended system matrix (SM_AB) for generating eigen-reconstruction image data ce(r); ande) subtracting the eigen-reconstruction image data ce(r) or weighted eigen-reconstruction image data ce′(r) from the input image data cin(r) for generating updated input image data cinn′(r).