MRI Image Correction Using Simulated Bloch Model Reference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Magnetic Resonance Imaging (MRI) systems face distortions and artifacts in reconstructed images due to geometric correlations and noise amplification, which existing methods struggle to effectively address.
Innovation Solution
Formulating image reconstruction or correction as an inverse problem using a simulated MRI image calculated from a B0 map, relaxation maps, and pulse sequence commands within a Bloch equation model, allowing for optimization of a cost function and regularization term to produce artifact-free images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If parallel imaging reconstruction is used to accelerate image acquisition, then scanning speed is improved, but geometric correlations cause noise amplification and image quality degradation
Solution Approach 1:
The method performs preliminary action by acquiring reference scan data before the actual imaging sequence and using it to calculate coil sensitivity maps and regularization parameters. This preliminary preparation enables the reconstruction algorithm to compensate for geometric correlations and noise amplification effects that would otherwise degrade image quality in parallel imaging.
Solution Approach 2:
The method implements feedback by using the reference scan data to inform and adjust the reconstruction process. The calculated coil sensitivity maps and regularization parameters from the reference scan provide feedback that guides the reconstruction algorithm to correct distortions and reduce noise amplification in the final image.
2Manufacturing precision
If regularization techniques are applied to reduce noise amplification, then image quality is improved, but the complexity of the reconstruction algorithm increases
Solution Approach 1:
The regularization parameters are calculated in advance from the reference scan data before the actual reconstruction. This preliminary calculation of regularization parameters simplifies the main reconstruction process by pre-determining the optimal regularization strength needed to reduce noise amplification without requiring complex iterative optimization during the main reconstruction.
Solution Approach 2:
The reference scan data serves as an intermediary that bridges the gap between the simplified acquisition and the complex reconstruction needs. It provides pre-calculated coil sensitivity maps and regularization parameters that mediate between the parallel imaging acceleration and the image quality requirements, reducing the computational complexity of the final reconstruction.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach improves image reconstruction by providing a simulated MRI image that serves as a reference for optimization, reducing distortions and noise, and enabling improved image correction and reconstruction quality.
Implementation Method 1
The simulated magnetic resonance image is calculated using a B0 map, and at least one of a transverse relaxation map and a longitudinal relaxation map, preferably both a T1 map, a T2 map and/or a T2* map, a magnetization map, and pulse sequence commands as input to a Bloch equation model or solver
Implementation Method 2
By applying time dependent magnetic field gradients and radio frequency (RF) pulses various quantities or properties of the subject can be measured spatially using MM
Data Source
AI summary
Disclosed is a medical imaging system (100, 300). The execution of machine executable instructions (120) causes a processor (104) to: receive (200) measured magnetic resonance imaging data (122) descriptive of a first region of interest (307) of a subject (318); receive (202) a B0 map (124), a T1 map (126), a T2 map (128), and a magnetization map (130) each descriptive of a second region of interest (309) of the subject; receive (204) pulse sequence commands (132); calculate (206) a simulated magnetic resonance image (136) of an overlapping region of interest (311) using at least the B0 map, the T1 map, the T2 map, the magnetization map, and the pulse sequence commands as input to a Bloch equation model (134); and reconstruct (208) a corrected magnetic resonance image from the measured magnetic resonance imaging data for the overlapping region of interest by solving an inverse problem. The inverse problem comprises an optimization of a cost function and a regularization term formed from the simulated magnetic resonance image.