| CPC G06T 11/005 (2013.01) [G06F 18/214 (2023.01); G06T 7/0014 (2013.01); G06T 2207/10088 (2013.01); G06T 2207/20084 (2013.01)] | 14 Claims |

|
1. A computer-implemented method for magnetic resonance imaging (MRI) reconstruction, comprising executing on a processor with steps of:
obtaining training data, and dividing the training data into first anatomical sections, wherein the training data comprise MRI images and MRI k-space data for training;
training anatomical-specific reconstruction models to obtain predicted MRI images from the MRI k-space data on each of the first anatomical sections;
acquiring scan data of an object while scanning the object, identifying second anatomical sections on the scan data, and determining reconstruction models corresponding to the second anatomical sections from the trained anatomical-specific reconstruction models;
reconstructing and obtaining reconstructed MRI images by using the determined reconstruction models and the predicted MRI images on each of the second anatomical sections; and
merging the reconstructed MRI images from each of the second anatomical sections to obtain at least one target MRI image, and outputting the at least one target MRI image;
wherein the MRI k-space data are three dimensional (3D) data with readout (kx) directions fully sampled, and each of the first anatomical sections and the second anatomical sections comprises at least one ky-kz plane perpendicular to a readout direction.
|