Reconstruction augmentation by constraining with intensity gradients in MRI
Inventors
Kermani, Ali Pouryazdanpanah • Afacan, Onur • Warfield, Simon K.
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12396654-B2
Publication Date
2025-08-26
Expiration Date
Abstract
A method for imaging a tissue region of a patient includes receiving a plurality of magnetic resonance (MR) signals of the tissue region, each MR signal being measured by a corresponding one of a plurality of coils, each coil having a different spatial sensitivity profile; and generating an image of the tissue region based on the plurality of MR signals and the spatial sensitivity profile of each coil, including applying one or more constraints to the image generation, wherein at least one of the constraints couples a first image value to one or more neighboring image values.
Core Innovation
The disclosed invention relates to parallel imaging in magnetic resonance imaging by receiving a plurality of magnetic resonance (MR) signals of a tissue region of a patient, where each MR signal is measured by a corresponding one of a plurality of coils, and each coil has a different spatial sensitivity profile. An image comprising a plurality of image values of the tissue region is generated based on the plurality of MR signals and the spatial sensitivity profile of each coil. During image generation, one or more constraints are applied so that at least one constraint couples a signal intensity or a spatial sensitivity profile associated with a first image value to one or more other image values.
The approach mitigates undersampling-related geometry-factor artifacts by enforcing equality constraints that couple neighboring spatial locations in voxel/pixel intensity or coil sensitivity-related quantities. The constraints are derived from spatial derivatives of an MRI encoding equation, producing linear equality constraints, and are incorporated into constrained reconstruction models solved using least-squares, penalized likelihood, and augmented Lagrangian formulations, including constrained least squares estimators and pseudo-inverse least squares formulations.
The disclosed model supports first-order, second-order, and combined derivative constraint formulations, and expresses constraints using spatial derivative operators. The disclosure further describes optional extensions of the constraint formulation, including filters or other operators, k-space implementations, and constraints involving motion and motion-transformation, as well as constraints between different coil subsets and between different contrasts. The document reports improved reconstruction quality and reduced geometry factor compared with GRAPPA across multiple two-dimensional and three-dimensional datasets and acceleration factors.
Claims Coverage
The independent claims are directed to a method for imaging a patient tissue region using constrained parallel-imaging reconstruction, a computing system, and a non-transitory computer readable medium for performing the same constrained image generation. Across the independent claims, the coverage centers on applying image-generation constraints that couple signal intensity or spatial sensitivity profile values between a first image value and one or more other image values, with further optional dependents specifying additional ways the coupling is expressed.
Constrained parallel-imaging image generation with coupling constraints
Applying one or more constraints to generate an image, wherein at least one constraint applies a coupling between a signal intensity or a spatial sensitivity profile of a first image value and one or more other image values of the plurality of image values.
Constrained coupling between image values derived from different coil subsets
Applying a constraint that couples a signal intensity or a spatial sensitivity profile value of a first image value associated with MR signals measured by a first subset of the plurality of coils to one or more other image values associated with MR signals measured by a second subset of the plurality of coils.
Computing system configured for constrained coupling-based image generation
A computing system that receives a plurality of MR signals measured by a plurality of coils with different spatial sensitivity profiles and generates an image based on the MR signals and spatial sensitivity profiles by applying one or more constraints, wherein at least one constraint applies coupling between a signal intensity or spatial sensitivity profile of a first image value to one or more other image values.
Non-transitory computer readable medium causing constrained coupling-based image generation
A non-transitory computer readable medium storing instructions that cause a computing system to receive a plurality of MR signals measured by a plurality of coils with different spatial sensitivity profiles and to generate an image based on the MR signals and spatial sensitivity profiles by applying one or more constraints, wherein at least one constraint applies coupling between a signal intensity or spatial sensitivity profile of a first image value to one or more other image values.
The independent claims consistently cover constrained parallel-imaging reconstruction in which at least one constraint couples signal intensity and/or spatial sensitivity profile values between image values during image generation, with the dependent coverage further specifying coupling between values derived from different coil subsets. The method claim, the computing system claim, and the computer-readable medium claim differ only by the recited implementation context while retaining the same coupling-based constraint concept.
Stated Advantages
Mitigates geometry-factor artifacts from undersampling by enforcing constraints that couple neighboring spatial locations.
Improves reconstruction/NRMSE and reduces geometry factor compared with GRAPPA across multiple 2D/3D datasets and acceleration factors.
Documented Applications
Parallel imaging reconstruction for magnetic resonance imaging, including constrained reconstruction models applied to multiple 2D/3D datasets and acceleration factors, with geometry factor comparisons against GRAPPA.
Constrained imaging with optional extensions including motion and motion-transformation constraints, and constraints between different coil subsets and between different contrasts.
Interested in licensing this patent?