Magnetic Susceptibility Mapping From Multiecho MRI Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional MRI methods struggle to accurately map magnetic susceptibility in brain tissue, particularly in the presence of nonlinear signal models and complex tissue compositions, leading to challenges in diagnosing neurological disorders and monitoring inflammation.
Innovation Solution
A system and method for collecting and processing multiecho complex MRI signals using a deep neural network to decompose magnetic susceptibility sources into components, generating high-quality susceptibility-based images by combining magnitude and phase information, and employing a digital twin for precise signal modeling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional MRI methods are used to map magnetic susceptibility, then the imaging process is simple and accessible, but the mapping accuracy is poor due to nonlinear signal models and complex tissue compositions
Solution Approach 1:
The patent segments the complex magnetic susceptibility mapping problem into multiple manageable components by dividing tissue susceptibility sources into distinct compartments (e.g., vasculature, myelin, diffuse background). This allows each compartment to be modeled and processed separately using simplified linear models, thereby improving overall mapping accuracy while managing computational complexity through structured decomposition.
Solution Approach 2:
The patent introduces an intermediary deep neural network that mediates between the complex physical signal model and the final susceptibility map. The neural network processes multiecho complex MRI signals, learns the nonlinear relationships between signal components and tissue properties, and outputs accurate susceptibility maps, effectively bridging the gap between simple imaging acquisition and accurate quantitative mapping.
2Measurement precision
If multiecho complex MRI signals are processed using deep neural networks, then mapping accuracy improves, but processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary actions by acquiring and processing multiple echo signals at different echo times before final susceptibility mapping. The deep neural network is trained on simulated multiecho data in advance, learning to extract susceptibility information efficiently. During actual processing, the pre-trained network can rapidly infer susceptibility maps from new multiecho data, reducing real-time computational burden while maintaining high accuracy.
3Reliability
If tissue composition complexity is accounted for in the signal model, then diagnostic accuracy improves, but the inverse problem becomes nonconvex and challenging to solve
Solution Approach 1:
The patent segments the tissue composition into distinct susceptibility compartments (vasculature, myelin, diffuse background), each with characteristic magnetic properties. By modeling each compartment separately with linear relationships and then combining them, the patent transforms the originally nonconvex inverse problem into a series of convex subproblems that are easier to solve while still capturing the full complexity of tissue composition for improved diagnostic accuracy.
Solution Approach 2:
The patent uses simulated multiecho complex MRI signals as copies or surrogates of actual patient data for training the deep neural network. These simulated datasets capture the full range of tissue compositions and susceptibility configurations, allowing the network to learn robust mappings from signals to susceptibility maps. This copying approach enables the network to handle complex tissue compositions effectively without requiring real-time solution of nonconvex inverse problems.
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
Enables accurate separation and mapping of tissue magnetic susceptibility sources, improving diagnostic accuracy and monitoring neurological disorders by providing high-quality susceptibility-based images.
Implementation Method 1
magnetic resonance imaging (MRI)... multiecho complex magnetic resonance susceptibility imaging data
Implementation Method 2
magnetic susceptibility affects the complex MRI signal in a nonlinear manner... inhomogeneity susceptibility field
Data Source
AI summary
Exemplary methods, systems and computer-accessible medium can be provided to generate images of tissue magnetism property from complex magnetic resonance imaging data using the Bayesian inference approach. The tissue magnetic susceptibility sources are organized into multiple components that differentially affect magnetic resonance susceptibility imaging signal, which is utilized to determine these susceptibility components. Exemplary methods, systems and computer accessible medium further enables susceptibility source determination. Thus, according to the exemplary embodiment, system, method and computer-accessible medium can be provided for determining magnetic susceptibility information and other tissue properties associated with at least one structure.