Iterative MR Reconstruction Using Offset Slices and Cost Functions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for creating quantitative MR images face challenges of long measurement times and limited spatial resolution, making them inefficient for clinical use.
Innovation Solution
A method that records MR data in an undersampled raw data space and uses model-based reconstruction, combining offset 2D slices with a cost function that accounts for slice displacement to achieve high-resolution images, reducing measurement time and improving spatial resolution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple data points are acquired at different echo times to model exponential signal evolution for quantitative determination, then measurement accuracy of T2 time is improved, but acquisition time increases significantly
Solution Approach 1:
The patent applies partial action by acquiring only a subset of the full k-space data (undersampling) rather than complete data at all echo times. This reduces the number of measurements needed while still enabling quantitative determination through iterative reconstruction that combines the undersampled data with signal evolution models to estimate the missing information.
Solution Approach 2:
The patent uses preliminary action by pre-defining signal evolution models (exponential decay models for T2, T1) that describe how MR signals change over time. These models are incorporated into the iterative reconstruction process to predict and fill in missing data points, allowing quantitative parameters to be determined from fewer actual measurements.
2Manufacturing precision
If slice thickness is reduced to improve spatial resolution in 2D imaging, then resolution in the slice direction is improved, but signal intensity decreases and noise increases
Solution Approach 1:
The patent transitions from 2D slice-based acquisition to 3D volumetric acquisition by sampling k-space in three dimensions. This allows thin slices to be combined into a volume, where the extended z-coverage provides increased signal through coherent summation, while maintaining high spatial resolution in all three dimensions through appropriate k-space sampling strategies.
Solution Approach 2:
The patent merges multiple thin slices into a unified 3D volume reconstruction. By combining the signal information from multiple closely-spaced thin slices in the iterative reconstruction process, the signal-to-noise ratio is improved through signal summation while maintaining the high spatial resolution of individual thin slices.
3Reliability
If 3D imaging techniques are used to increase signal-to-noise ratio, then spatial resolution is improved, but specific absorption rate increases excessively
Solution Approach 1:
The patent uses periodic action by implementing variable flip angle sequences where the RF pulse flip angles are modulated according to a periodic pattern (e.g., alternating between higher and lower angles). This periodic variation allows the sequence to maintain signal-to-noise ratio while distributing the SAR burden over time, preventing excessive localized heating.
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting RF pulse parameters (flip angles, repetition times, echo times) and gradient parameters during the 3D acquisition sequence. These parameter modifications enable the sequence to achieve the desired signal-to-noise ratio and spatial resolution while keeping the specific absorption rate within safety limits through optimized RF energy deposition.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention comprises a method for generating quantitative MR images of a subject. A first MR dataset of the subject is acquired in an undersampled raw data space, wherein the subject is acquired in several 2-dimensional slices, the resolution in one layer of each slice being higher than perpendicular to the layer plane, and the multiple 2-dimensional slices being offset from each other by a distance that is less than the resolution perpendicular to the layer plane. Further MR raw data points of the first MR dataset are reconstructed using a model with a cost function that is minimized. The cost function takes into account the offset of the multiple 2-dimensional slices perpendicular to the layer plane.