Trimmed Autocalibrating K-space Estimation for Motion-Corrupted MRI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing parallel MRI techniques face challenges in accurately reconstructing images from undersampled k-space data, particularly due to motion corruption, which leads to errors and inconsistencies, and existing methods like IRLS are insufficient in handling gross errors and outliers, resulting in loss of precision and sensitivity issues at different spatial frequencies.
Innovation Solution
The Trimmed Autocalibrating and K-space Estimation (TAKE) method employs structured low-rank matrix completion and iterative rank reduction to handle motion-corrupted data, using noise variance estimation, minimum variance filtering, robust location parameters, trimming outliers, and incorporating additional constraints like conjugate symmetry to reconstruct images robustly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If parallel imaging techniques are used to reduce scanning time by undersampling k-space, then productivity is improved, but measurement precision deteriorates due to motion corruption and data inconsistencies
Solution Approach 1:
The patent extracts and removes corrupted data points from the k-space matrix by calculating residuals between observed and predicted values. Data points with residuals exceeding a threshold are identified as outliers and removed, thereby eliminating motion-corrupted information that would otherwise degrade reconstruction accuracy while preserving the accelerated scanning benefits.
Solution Approach 2:
The patent changes the approach from direct image reconstruction to a two-stage process: first estimating the noise variance and removing outliers based on residual analysis, then performing reconstruction. This parameter-based filtering approach adapts to the specific noise characteristics of each scan, improving measurement precision without sacrificing the productivity gains from undersampling.
2Reliability
If iterative reweighted least squares (IRLS) is used to handle outliers, then reliability is improved, but manufacturing precision deteriorates due to loss of precision and sensitivity issues at different spatial frequencies
Solution Approach 1:
The patent replaces the IRLS iterative optimization mechanism with a direct residual-based outlier detection and removal approach. Instead of iteratively reweighting all data points, the method calculates residuals, identifies outliers exceeding a threshold, and removes them directly. This substitution maintains reliability by effectively handling outliers while preserving manufacturing precision by avoiding the precision loss and sensitivity issues inherent in IRLS.
3Measurement precision
If a separate calibration data acquisition is conducted to obtain coil sensitivity profiles, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent merges the calibration data acquisition with the diagnostic data acquisition by using the same undersampled k-space data for both purposes. The autocalibration signals are embedded within the diagnostic scan itself, allowing the system to estimate coil sensitivity profiles without requiring a separate calibration scan. This eliminates the time loss while maintaining sufficient measurement precision through the combined use of autocalibration and outlier removal techniques.
4Productivity
If excessive undersampling is applied to reduce scan time, then productivity is improved, but reliability deteriorates due to loss of calibration data
Solution Approach 1:
The patent applies partial action by using only a subset of the undersampled k-space data for calibration purposes, specifically targeting regions that provide sufficient information for sensitivity profile estimation. By selectively using portions of the undersampled data for calibration and applying outlier removal to the remaining data, the system maintains reliability even with excessive undersampling that would otherwise completely lose calibration information.
Data Source
AI summary
In a method and magnetic resonance (MR) apparatus for MR image reconstruction, an image reconstruction algorithm is used that operates on one calibration matrix that is formed by reorganizing a number of individual, undersampled k-space data sets respectively acquired by multiple reception coils in a parallel MR data acquisition from a subject exhibiting motion. The motion causes the k-space data sets to exhibit errors. In order to use the calibration matrix in the reconstruction algorithm, it is subjected to an iterative rank reduction procedure in which, in each iteration, a residual is calculated for each data point that represents how poorly, due to motion-induced corruptions, that data point satisfies the low rank constraint, and non-satisfying data points are removed from the data point for the next iteration. The resulting low rank matrix at the end of the iterations is then used to produce images with fewer motion-induced errors.