Randomized MR Data Acquisition for Motion Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for recording magnetic resonance (MR) signals from moving examination objects result in incorrect data registration and image artifacts due to movement, such as breathing or heart movement, and current solutions either limit measurement time, increase complexity, or reduce spatial resolution.
Innovation Solution
A method where movement information of the examination object is detected and grouped into phases, with a temporally randomly distributed sequence for recording raw data lines to fill the raw data space, ensuring a statistically random distribution of data, allowing for effective reconstruction using Compressed Sensing Technology and minimizing movement artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If segmented recording technique is used to divide data acquisition into multiple movement phases, then movement artifacts are reduced, but device complexity increases and measurement time increases
Solution Approach 1:
The patent applies dynamics by using a random sampling pattern that adapts to movement phases rather than rigid segmentation. The sampling trajectory randomly selects positions in k-space based on detected movement phase, allowing flexible data acquisition that accommodates physiological motion without requiring complex phase-based segmentation protocols.
Solution Approach 2:
The patent changes the sampling parameter from fixed grid-based acquisition to random sampling with variable density. By adjusting the sampling probability and pattern based on movement detection, the system optimizes data collection efficiency while maintaining image quality, avoiding the complexity of segmented recording.
2Loss of time
If Compressed Sensing Technology is used with severe under-sampling of raw data spaces, then measurement time is reduced, but manufacturing precision of raw data decreases
Solution Approach 1:
The patent applies local quality by using variable density sampling where different regions of k-space are sampled at different rates. High-frequency regions are sampled sparsely while low-frequency regions receive denser sampling, optimizing the balance between measurement time and raw data quality for compressed sensing reconstruction.
Solution Approach 2:
The patent performs preliminary movement detection and phase identification before data acquisition. This allows the sampling pattern to be pre-adjusted based on the detected movement state, ensuring optimal sampling coverage is achieved before the actual measurement begins, thereby maintaining raw data quality even with reduced sampling.
3Measurement precision
If peripheral raw data are recorded to allow movement correction, then movement artifacts are reduced, but spatial resolution decreases due to missing central raw data portions
Solution Approach 1:
The patent applies asymmetry by deliberately sampling peripheral k-space regions more frequently than central regions in certain movement phases. This asymmetric sampling strategy captures movement-related information from peripheral areas while using compressed sensing to reconstruct the missing central data, maintaining both movement correction capability and spatial resolution.
Solution Approach 2:
The patent uses compressed sensing reconstruction algorithms as an intermediary to fill in missing central k-space data based on the randomly sampled peripheral data and detected movement patterns. This intermediary process recovers the spatial resolution information that would otherwise be lost, while preserving the movement correction benefits of peripheral sampling.
Data Source
AI summary
In a method and apparatus for recording magnetic resonance (MR) signals from an examination object, raw data space is filled with MR signals in raw data lines. Movement information of the examination object is detected during recording of the MR signals and the movement information is grouped into different movement phases of the examination object. A temporally randomly distributed sequence of the recording of the raw data lines is determined, with which at least one predetermined portion of the raw data space is filled MR signals. The MR signals are acquired in the determined temporally randomly distributed sequence of the raw data lines in the predetermined portion. Each recorded raw data line is allocated to one of the movement phases of the examination object.