EPI MRI Phase Correction Using Dual-Polarity GRAPPA Kernels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing EPI methods struggle to effectively correct high-order phase errors caused by gradient hardware defects and eddy effects, leading to residual ghosts in MRI images, and are difficult to integrate with other advanced imaging reconstruction methods.
Innovation Solution
A method and device that decouples phase error correction and missing k-space line estimation by using dual-polarity GRAPPA kernels to fit and correct low-order and high-order phase errors in EPI data, enabling integration with parallel reconstruction and deep learning methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the traditional NGC method is used to correct phase errors, then zero-order and first-order phase errors can be corrected, but high-order phase errors caused by space-related eddy effects cannot be effectively eliminated, resulting in residual ghosts
Solution Approach 1:
The patent changes the parameters of phase error correction by using DPG kernels that can model and correct high-order phase errors beyond the traditional zero-order and first-order corrections. This involves adjusting the correction model to include higher-order terms that account for space-related eddy effects, thereby eliminating residual ghosts while maintaining correction accuracy.
Solution Approach 2:
The patent combines multiple correction approaches by integrating DPG kernel-based high-order phase error correction with existing GRAPPA reconstruction methods. This composite approach merges the strengths of different correction techniques to achieve both low-order and high-order phase error correction, resolving the limitation of traditional single-method corrections.
2Reliability
If the DPG method is used to correct residual ghosts in accelerated EPI data, then phase errors can be corrected, but the method is difficult to integrate with other accelerated imaging reconstruction methods like sensitivity encoding and deep learning
Solution Approach 1:
The patent extracts the phase error correction function from the integrated DPG-GRAPPA framework and formulates it as a separate, independent correction step. This allows the correction method to be applied universally across different reconstruction approaches (sensitivity encoding, deep learning, etc.) without requiring method-specific integration, thereby improving adaptability while maintaining correction effectiveness.
Solution Approach 2:
The patent creates a universal phase error correction framework using DPG kernels that can be applied across multiple reconstruction methods. The correction approach is designed to be method-agnostic, working with GRAPPA, sensitivity encoding, deep learning, and other accelerated imaging techniques, thus achieving multi-functionality and broad adaptability.
3Measurement precision
If EPI data is divided into sub-frames for GESTE method, then phase errors can be corrected, but the effective acceleration factor is doubled, reducing productivity
Solution Approach 1:
The patent applies phase error correction using DPG kernels as a preliminary step before reconstruction, working on the raw k-space data directly. This preliminary correction eliminates the need for subsequent division into sub-frames, maintaining the original acceleration factor while achieving accurate phase error correction, thus preserving productivity.
Data Source
AI summary
Techniques are provided for performing echo planar imaging (EPI) data correction. This includes obtaining positive and negative readout gradient calibration data of an imaging target through non-accelerated EPI acquisitions; respectively adopting first and second DPG kernels to be fitted and respectively used to eliminate phase errors of positive and negative readout gradients to fit the positive and negative readout gradient calibration data of the imaging target, with the fitting targets being positive and negative readout gradient data, respectively, in ghost-free target ACS data, and obtaining, after the fitting, a first and a second DPG kernel for final use; obtaining imaging data of the imaging target through an EPI acquisition; adopting the first and second DPG kernels to correct the phase errors of the imaging data to obtain phase-error-free imaging data.