Magnetic resonance scan calibration and reconstruction technique for multi-shot, multi-echo imaging

In a method of magnetic resonance imaging in which magnetic resonance is excited in dipoles which are induced to form echoes at each of a plurality of data collection intervals following each resonance excitation, the echoes being read out along a read axis in the presence of a read gradient to form a series of data lines for reconstruction into an output image representation, the improvement comprising the steps of:

• (a) receiving a plurality of data lines, each collected at one of the data collection intervals;
• (b) generating data collection matrices for each data collection interval;
• (b) generating a plurality of two-dimensional phase-correction matrices, each phase-correction matrix corresponding to one of the data collection intervals;
• (c) reconstructing an incomplete set of data lines collected during each of the data collection intervals into a corresponding partial image;
• (d) multiplying each partial image by the phase-correction matrix corresponding to the same data collection interval to generate a plurality of two-dimensionally phase-corrected partial images;
• (e) adding the two-dimensionally phase-corrected partial images to generate a composite image.

A method for magnetic resonance imaging comprising:

• refocusing magnetic resonance signals to generate echoes;
• phase-encoding the echoes to generate data lines tracing along each of a plurality of segments of k-space;
• sorting the data lines among the plurality of segments of k-space;
• generating phase-correction matrices from the data lines, each matrix corresponding to one of the segments of k-space;
• generating a plurality of partial images from the data lines, each partial image corresponding to one of the segments of k-space;
• generating a plurality of phase-corrected partial images from the plurality of partial images and phase-correction matrices, each phase-corrected partial image corresponding to one of the segments of k-space;
• and,
• summing the phase-corrected partial images to generate a composite image.

A method of partial Fourier imaging comprising:

• (a) inversely Fourier transforming a composite image in at least a column direction to generate a set of data lines, which set of data lines includes data lines from a central segment of k-space and a second segment of k-space to one side of the central segment;
• (b) zero-filling a third segment of k-space to the opposite side of the central segment;
• (c) filtering the data lines to extract the data lines from the central segment of k-space, Fourier transforming the extracted data lines to generate a low resolution image, and normalizing the low resolution image to generate a low resolution phase map;
• (d) filtering the data lines to extract the data lines from the central and second segments of k-space;
• (e) Fourier transforming the extracted data lines to generate an image, taking an absolute value of the image to generate a magnitude image, and combining the magnitude image with the low resolution phase map to generate a complex image;
• (f) inversely Fourier transforming the complex image in at least the column direction to generate data lines in the central, second, and third segments of k-space, and filtering the data lines to extract the data lines from the third segment of k-space;
• (g) combining the extracted data lines from the central and second segments of k-space with the data extracted from the third segment of k-space and repeating step (e) and step (f) such that the generated complex image is iteratively improved.