Magnetic Resonance Fingerprinting Neural Network Scheduling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current magnetic resonance fingerprinting (MRF) techniques require lengthy acquisition times due to the need for a large number of repetitions and extensive computational resources, limiting their clinical usage and introducing errors from undersampling.
Innovation Solution
The method employs a neural network to optimize acquisition parameters, reducing the number of repetition time periods and enhancing discrimination between quantitative parameters, allowing for faster data acquisition and reduced computational burden by selecting a sparse subset of acquisition parameters and using a compact representation of the acquisition process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large number of TR periods are used for MRF acquisitions, then parameter estimation accuracy is improved, but acquisition time increases significantly
Solution Approach 1:
The patent applies parameter changes by optimizing acquisition parameters (flip angle, TR, TE, bandwidth) to maximize signal discrimination between different tissue parameters. This allows achieving accurate parameter estimation with fewer TR periods by selecting parameters that create more distinctive signal evolutions, directly resolving the contradiction between accuracy and acquisition time
Solution Approach 2:
The patent performs preliminary optimization of acquisition parameters before the actual MRF acquisition. By pre-calculating the optimal parameter schedule that maximizes discrimination, the system can achieve accurate parameter estimation in fewer TR periods, thereby reducing total acquisition time while maintaining precision
2Productivity
If highly undersampled spiral k-space acquisition is used, then acquisition speed is improved, but image quality and parameter estimation accuracy deteriorate
Solution Approach 1:
The patent optimizes sampling parameters including spiral trajectory, sampling density, and k-space coverage to achieve the best balance between undersampling speed and parameter estimation accuracy. By carefully selecting sampling parameters, the system maintains acceptable image quality and accuracy while significantly reducing acquisition time
3Measurement precision
If spiral trajectory is changed from one time point to the next, then signal discrimination is improved, but computational complexity and data processing burden increase
Solution Approach 1:
The patent systematically varies spiral trajectory parameters (radius, orientation, sampling density) across different TR periods to maximize signal discrimination. This structured parameter variation achieves better parameter estimation accuracy while the variations follow predictable patterns that facilitate computational processing
4Measurement precision
If approximately 1000-2000 time points are acquired per slice, then parameter estimation accuracy is improved, but total acquisition time for volumetric imaging becomes excessively long
Solution Approach 1:
The patent optimizes the number of time points per slice by adjusting acquisition parameters to maximize discrimination efficiency. This allows reducing the number of required time points while maintaining parameter estimation accuracy, thereby significantly reducing total acquisition time for volumetric imaging
Solution Approach 2:
The patent applies partial action by acquiring data from only a selected subset of slices or regions of interest rather than all slices, using optimized parameters to achieve sufficient accuracy with fewer measurements, thus reducing total acquisition time while maintaining acceptable precision
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 significantly reduces acquisition time, minimizes errors, and improves the efficiency of parameter estimation while maintaining accuracy, enabling more widespread clinical application of MRF techniques.
Implementation Method 1
Magnetic resonance fingerprinting (MRF) is an technique that enables quantitative mapping of tissue or other material properties based on varied measurements
Implementation Method 2
The computer system is programmed to use a neural network to estimate acquisition parameters that are selected to direct a magnetic resonance system
Implementation Method 3
Quantitative parameters of the subject are then estimated by comparing the acquired data with a dictionary database comprising a plurality of different signal templates
Data Source
AI summary
A system and method is provided for estimating quantitative parameters of a subject using a magnetic resonance system. The method includes using a neural network, estimating acquisition parameters that are selected to direct a magnetic resonance system to generate a plurality of different signal evolutions that elicit discrimination between different quantitative parameters in a desired number of repetition time (TR) periods. The method also includes acquiring data with the magnetic resonance system by performing a plurality of pulse sequences using the estimated acquisition parameters, where the acquired data representing the plurality of different signal evolutions that elicit discrimination between different quantitative parameters. The method further includes estimating quantitative parameters of the subject by comparing the acquired data with a dictionary database comprising a plurality of different signal templates and generating a report indicating the quantitative parameters.