Magnetic Resonance Parameter Map Reconstruction via GRAPPA and MARTINI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining parameter maps in magnetic resonance imaging are time-consuming and susceptible to errors due to violations of the signal model, particularly in the presence of blood flow, partial volume effects, head motion, and noise, which limits the acceleration of measurement times and image quality.
Innovation Solution
A method combining parallel imaging techniques with model-based approaches, specifically using GRAPPA and MARTINI, to undersample and interpolate magnetic resonance data, reducing artifacts and enhancing robustness, allowing for faster measurement times and improved image quality by interpolating missing data and using redundancy information from multiple receive coils.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If model-based approaches with undersampling of the k-space are used to accelerate measurement, then measurement time is reduced, but the methods become susceptible to errors due to violations of the signal model (blood flow, partial volume effects, head motion, noise)
Solution Approach 1:
The k-space is divided into multiple portions, and only a subset of portions is sampled for each echo time. This segmentation allows undersampling while maintaining enough sampled data to constrain the optimization problem, reducing measurement time while preserving reliability through the iterative reconstruction process that uses the signal model.
Solution Approach 2:
Instead of sampling the entire k-space for each echo time, only a partial subset of k-space portions is sampled. This partial action reduces the amount of data acquisition needed, thereby reducing measurement time, while the iterative optimization with signal model constraints ensures that the parameter map remains reliable despite the reduced sampling.
2Loss of time
If highly undersampled data is used to achieve faster measurement, then measurement time is reduced, but data quality and image quality deteriorate with increased artifacts and noise
Solution Approach 1:
An iterative optimization method is employed where the parameter map is repeatedly refined by comparing hypothesis data (generated from the signal model using current parameter estimates) with the actually measured undersampled magnetic resonance data. This feedback loop continues until convergence, ensuring that the final parameter map accurately represents the underlying tissue properties despite the undersampled input data, thereby maintaining data quality while achieving fast measurement.
Solution Approach 2:
A signal model describing the magnetization is established before the actual parameter estimation. This preliminary action provides a theoretical framework that guides the iterative reconstruction process, allowing the system to infer missing information from the undersampled data and maintain high data quality even with reduced sampling.
3Manufacturing precision
If conventional fully-sampled k-space acquisition is used, then data quality is maintained, but measurement time becomes excessively long
Solution Approach 1:
The k-space is segmented into multiple portions that are sampled at different echo times rather than fully sampling all portions for each echo time. This segmentation strategy reduces the total number of measurements required, thereby reducing measurement time, while the iterative reconstruction process ensures that data quality is maintained by utilizing the signal model to fill in missing information.
Solution Approach 2:
Instead of performing the complete k-space sampling for each echo time, a partial subset of k-space portions is sampled. This partial action significantly reduces measurement time while the model-based iterative reconstruction compensates for the missing data, maintaining data quality equivalent to conventional methods.
4Loss of time
If parallel imaging techniques are combined with model-based approaches, then measurement time is accelerated, but device complexity increases
Solution Approach 1:
Parallel imaging techniques are merged with model-based iterative reconstruction approaches. The parallel imaging provides accelerated data acquisition by undersampling, while the model-based approach provides the reconstruction framework. This combination achieves significant measurement time acceleration, and the integrated nature of the approach manages complexity by unifying the acquisition and reconstruction processes under a common optimization framework.
Data Source
AI summary
A method is disclosed for recording a parameter map of a target region via a magnetic resonance device. In at least one embodiment, an optimization method is used for the iterative reconstruction of the parameter map. In the optimization method, the deviation of undersampled magnetic resonance data of the target region present in the k-space for different echo times, magnetic resonance data of a portion of the k-space being present in each case for each echo time, is assessed from hypothesis data of a current hypothesis for the parameter map obtained as a function of the parameter from a model for the magnetization. To determine the magnetic resonance data of a portion of the k-space, undersampled raw data is initially acquired within the portions by way of the magnetic resonance device embodied for parallel imaging, and missing magnetic resonance data within the portions is completed by way of interpolation.