MR Image Combination via Cross-Calibration Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The generation of combination image datasets using parallel imaging techniques often results in artifacts such as third arm artifacts, spoiling, and foldover artifacts, which degrade image quality.
Innovation Solution
The method involves reconstructing intermediate image datasets using calibration data from associated and different raw datasets recorded with the same coil, and combining these datasets using addition or subtraction to generate a combination image dataset, while utilizing GRAPPA or Auto-SMASH-based methods for improved SNR and artifact reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If parallel imaging techniques (PPA) are used to shorten measurement time by recording reduced datasets with coil arrays, then productivity is improved, but manufacturing precision deteriorates due to aliasing artifacts and foldover effects
Solution Approach 1:
The patent introduces an intermediate reconstruction step that processes the reduced datasets from parallel imaging before final combination. GRAPPA and Auto-SMASH methods act as intermediary techniques that reconstruct missing k-space lines and unfold aliasing artifacts, thereby maintaining image quality while preserving the speed benefits of parallel imaging
Solution Approach 2:
The patent replaces traditional sequential imaging methods with parallel imaging techniques that use multiple coils simultaneously. This substitution enables faster data acquisition by recording multiple reduced datasets in parallel, trading off some direct image quality for improved productivity, which is then compensated through computational reconstruction
2Device complexity
If calibration data from only the associated raw dataset is used during reconstruction, then device complexity is reduced, but measurement precision deteriorates due to insufficient calibration information
Solution Approach 1:
The patent merges calibration data from multiple raw datasets recorded with the same coil into a combined calibration dataset. This combining process pools calibration information across different parameter sets (e.g., different echo times), providing more robust coil sensitivity estimates and improving measurement precision without significantly increasing reconstruction complexity
Solution Approach 2:
The patent creates a universal calibration dataset that serves multiple reconstruction purposes across different parameter sets. The combined calibration data from one coil can be used to reconstruct images from multiple raw datasets recorded with different parameters, making the calibration process more efficient and precise
Data Source
AI summary
In a method and apparatus for generating a magnetic resonance (MR) image MR data are acquired from a subject as datasets in parallel with multiple RF coils, with first parallel dataset being acquired with a first parameter set and at least one further parallel dataset being acquired with a second parameter set. A first intermediate image dataset and at least one further intermediate image dataset are reconstructed with at least one of (a) the first intermediate image dataset being reconstructed from said first parallel dataset using a calibration data item derived from said at least one further parameter set, and (b) said at least one further intermediate image dataset is reconstructed from said at least one further parallel dataset using a calibration data item derived from said first parameter set. A combination image dataset is generated by combining said first intermediate image dataset and said at least one further intermediate dataset.