Rigid Body Motion and Background Phase Correction Using GRAPPA Aided Motion Transformed Coil Sensitivity Maps

A method for magnetic resonance imaging (MRI) of a subject, the method comprising:

• (a) acquiring calibration data to establish initial coil sensitivity maps for a plurality of radiofrequency coils;
• (b) acquiring raw k-space data from the subject using an MRI scanner, wherein the raw k-space data is associated with a plurality of motion states of the subject during acquisition;
• (c) estimating motion parameters and background phase variations for each of the plurality of motion states from the acquired data;
• (d) generating a plurality of interpolation kernels by training a machine learning model to learn relationships in k-space introduced by the motion parameters and background phase variations;
• (e) applying the plurality of interpolation kernels to the raw k-space data to produce a cleaned k-space dataset, wherein applying the kernels aligns coil sensitivities with subject motion and rewinds background phase variations;
• and
• (f) reconstructing an image from the cleaned k-space dataset using an iterative parallel imaging reconstruction algorithm, wherein a singular forward model transformation is performed for the plurality of motion states during each iteration.