| CPC A61B 5/055 (2013.01) [G01R 33/3415 (2013.01); G01R 33/482 (2013.01); G01R 33/5611 (2013.01); G06T 5/70 (2024.01); G06T 2207/10088 (2013.01)] | 14 Claims |

|
1. A computer-implemented method of reconstructing magnetic resonance imaging (MRI) images, the method comprising:
for each coil of a receiver coil array of an MRI apparatus, determining a respective set of two or more coil sensitivity maps using an autocalibration protocol comprising an Eigenvalue-based Spatially Constrained Iterative Reconstruction Technique (ESPIRIT) autocalibration protocol densely sampling a part of k-space and determining a respective coil sensitivity data structure comprising the respective two or more coil sensitivity maps in a stitched arrangement along a predetermined direction in k-space, a stitching factor of the stitched arrangement corresponding to a number of the two or more coil sensitivity maps of the respective set,
for each coil of the receiver coil array, acquiring k-space data using an undersampling trajectory in k-space, and determining a respective measurement data structure comprising respective k-space data rearranged in k-space on an adjusted undersampling k-space trajectory that is determined by stretching the undersampling trajectory in k-space by the stitching factor along the predetermined direction, and
based on the coil sensitivity data structures determined for the coils of the receiver coil array, and further based on the respective measurement data structures determined for the coils of the receiver coil array, performing an iterative optimization based on a SENSitivity Encoding (SENSE) reconstruction algorithm to obtain at least one MRI image,
wherein the iterative optimization comprises, for at least a portion of iterations of multiple iterations of the iterative optimization, a regularization operation and a data-consistency operation, and
wherein the data-consistency operation is based on differences between the measurement data structures and respective synthesized measurement data structures, the synthesized measurement data structures being based on a k-space representation of at least one prior MRI image of the multiple iterations, the adjusted undersampling k-space trajectory, and the coil sensitivity data structures.
|