Magnetic Resonance Fingerprinting Dictionary Compression via SVD

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Magnetic Resonance Fingerprinting techniques face inefficiencies due to computationally expensive iterative processes for dictionary search and signal comparison, which hinder accurate and timely retrieval of magnetic resonance parameters from undersampled signal responses.

Innovation Solution

The implementation of an iterative gradient proximal process that incorporates spatial and low-rank regularization by compressing the fingerprint dictionary using Singular Value Decomposition, reducing the rank of the dictionary and utilizing autocalibration data to accelerate data fidelity and fingerprint matching, and applying spatial regularization to improve image reconstruction efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If iterative dictionary search and signal comparison processes are used to achieve accurate magnetic resonance parameter retrieval, then measurement precision is improved, but productivity deteriorates due to computationally expensive operations

Engineering Contradiction:

Improveaccuracy of magnetic resonance parameter retrievalVSAvoidcomputational efficiency

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent segments the dictionary search process by organizing the dictionary into multiple sub-dictionaries based on tissue types or parameter ranges. This allows the iterative algorithm to search smaller, more manageable subsets of the full dictionary, reducing computational complexity while maintaining accurate parameter retrieval through targeted comparisons

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary preprocessing of the signal response data before initiating the iterative dictionary search. This includes initial parameter estimation, signal normalization, and filtering operations that prepare the data in advance, allowing the iterative refinement process to converge faster with fewer computational iterations

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If full-rank fingerprint dictionary is used for comprehensive signal matching, then measurement precision is improved, but device complexity increases due to large data storage and processing requirements

Engineering Contradiction:

Improvesignal matching accuracyVSAvoiddata storage and processing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts and removes redundant or less informative dimensions from the fingerprint dictionary. By identifying and eliminating duplicate or highly correlated dictionary entries, the system maintains sufficient signal matching accuracy while significantly reducing the dictionary size and associated storage and processing complexity

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent transforms the dictionary representation by changing parameters such as compression ratios, dimensionality reduction levels, or basis function selections. These parameter adjustments allow the system to achieve an optimal balance between dictionary comprehensiveness for accurate matching and compactness for reduced complexity

Inventive Principle:

Principle #35Parameter changes

3Productivity

If undersampled signal response is used for faster acquisition, then productivity is improved, but measurement precision deteriorates due to insufficient data for accurate dictionary matching

Engineering Contradiction:

Improveacquisition speedVSAvoidparameter retrieval accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent implements feedback mechanisms where the iterative dictionary search process continuously refines parameter estimates based on matching errors. Each iteration uses the previous iteration's results to guide the next search, progressively improving accuracy even from undersampled data by leveraging the feedback from mismatched signals to correct parameter estimates

Inventive Principle:

Principle #23Feedback

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 approach enhances the accuracy and speed of magnetic resonance image reconstruction by reducing computational time and measurement errors, improving the sensitivity and specificity of MR studies, and enabling more efficient diagnostic methodologies.

Implementation Method 1

A compressed dictionary is multiplied by an additional compression matrix obtained via a Singular Value Decomposition algorithm applied on autocalibration data

Methodology Applied

Scientific EffectSingular Value Decomposition:

Implementation Method 2

Each simulated response stored in the dictionary is generated by running Bloch equations with relevant values for magnetic resonance parameters (T1, T2, PD, etc.)

Methodology Applied

Scientific EffectBloch equations:

Data Source

PatentUS10866298B2Low rank and spatial regularization model for magnetic resonance fingerprinting

Publication Date: 2020.12.15 SIEMENS HEALTHINEERS AG

AI summary

Systems and methods are provided for iterative reconstruction of a magnetic resonance image using Magnetic Resonance Fingerprinting (MRF). An image series is estimated according to the following three steps: a gradient step to improve data consistency, fingerprint matching, and a spatial regularization. Singular Value Decomposition (SVD) compression may be used along the time dimension to accelerate both the matching and the spatial regularization that operates in the compressed domain as well as to enforce low-rank regularization.