# US 2023/0055826 A1 · Systems and methods for magnetic resonance image reconstruction from incomplete k-space data

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2023/0055826 A1
- Document ID: US20230055826A1
- Status: Published application
- Kind: A1
- Application number: 17/797,119
- Filed: 2021-02-02
- Published: 2023-02-23
- Earliest claimed: 2020-02-04
- Assignees: The University of Hong Kong
- Applicants: The University of Hong Kong
- Inventors: Ed Xuekui Wu; Yilong Liu; Linfang Xiao
- Web page: https://patentgrep.com/patent/US20230055826A1

## CPC

- G01R 33/5608: {Data processing and visualization specially adapted for MR, e.g. for feature analysis and pattern recognition on the basis of measured MR data, segmentation of measured MR data, edge contour detection on the basis of measured MR data, for enhancing measured MR data in terms of signal-to-noise ratio by means of noise filtering or apodization, for enhancing measured MR data in terms of resolution by means for deblurring, windowing, zero filling, or generation of gray-scaled images, colour-coded images or images displaying vectors instead of pixels (image data processing or generation, in general G06T)}
- G01R 33/561: by reduction of the scanning time, i.e. fast acquiring systems, e.g. using echo-planar pulse sequences
- G01R: MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES (indicating correct tuning of resonant circuits H03J3/12)

## Abstract

Disclosed are deep learning based methods for magnetic resonance imaging (MRI) image reconstruction from partial Fourier-space (i.e., k-space) data, involving: obtaining high-quality complex MRI image data or fully-sampled k-space data as training data; training reconstruction models to predict high-quality complex MRI image data or complete k-space data from incomplete or partial k-space data; and applying trained models to reconstruct high-quality complex MRI image data or complete k-space data from partial k-space data.

## Claims

### Claim 1

A method for MRI reconstruction of partial Fourier k-space data using deep learning models, comprising:
obtaining large training data comprising complex MRI images or fully-sampled k-space data;
training artificial neural networks to predict residual complex images from incomplete k-space data; and
reconstructing high-quality MRI images from individual incomplete k-space data that are partially sampled and the residual complex images predicted by using the trained artificial neural networks.

### Claim 2 (depends on 1)

The method of claim 1, wherein the training data comprises 2D, 3D and multi-channel.

### Claim 3 (depends on 1)

The method of claim 1, wherein the complex MRI images or complete k-space data MRI k-space data comprises at least one of 2D, 3D and multi-channel.

### Claim 4 (depends on 1)

The method of claim 1, wherein the incomplete k-space data comprises at least one of 2D, 3D and multi-channel.

### Claim 5 (depends on 1)

The method of claim 1, wherein the incomplete k-space data comprises partially sampled in one or more directions during Cartesian k-space sampling via phase encoding or/and frequency encoding schemes.

### Claim 6 (depends on 1)

The method of claim 1, wherein the incomplete k-space data comprises partially sampled in angular direction during polar k-space sampling via projection encoding schemes.

### Claim 7 (depends on 1)

The method of claim 1, wherein the incomplete k-space data comprises partially sampled through non-Cartesian and non-polar k-space trajectories, such as during spiral sampling schemes.

### Claim 8 (depends on 1, 5, 6, 7)

The method of claim 1, wherein the corresponding training data to claims 5 to 7 above has the same but complete k-space sampling trajectories or patterns during MRI data acquisition or through post-acquisition retrospective k-space data rebinning.

### Claim 9

A method for MRI reconstruction of partial Fourier k-space data using deep learning models, comprising:
obtaining large training data comprising complex MRI images or fully-sampled k-space data;
training artificial neural networks to predict residual complex images from incomplete k-space data; and
reconstructing high-quality MRI images from individual incomplete k-space data that are partially sampled and the residual complex images predicted by using the trained artificial neural networks.

### Claim 10 (depends on 9)

The method of claim 9, wherein the artificial neural networks accommodate data with dimensions high than 2D (e.g., 3D and 4D MRI), and multi-channel nature (e.g., typical multi-receiving-channel data as in parallel imaging).

### Claim 11 (depends on 9)

The method of claim 9, wherein the artificial neural networks are based in complex image domain as shown in FIG. 1, k-space domain, or in between.

### Claim 12 (depends on 9)

The method of claim 9, wherein the artificial neural networks comprise a convolutional neural network (CNN).

### Claim 13 (depends on 9)

The method of claim 9, wherein the artificial neural networks comprise a generative adversarial network (GAN).

### Claim 14 (depends on 9)

The method of claim 9, wherein the artificial neural networks comprise an autoencoder network.

### Claim 15 (depends on 9)

The method of claim 9, wherein the artificial neural networks comprise other type of networks, modification, or combination of existing network architectures.

### Claim 16

A method for MRI reconstruction of partial Fourier k-space data using deep learning models, comprising:
obtaining large training data comprising complex MRI images or fully-sampled k-space data, wherein the training data (complete) and partial Fourier data (incomplete) are not required to be from the same or similar image orientations, objects (subjects included), anatomical structures (organs included), MRI contrasts, or MRI scanners or MRI field strength;
training artificial neural networks to predict residual complex images from incomplete k-space data; and
reconstructing high-quality MRI images from individual incomplete k-space data that are partially sampled and the residual complex images predicted by using the trained artificial neural networks.

### Claim 17 (depends on 16)

The method of claim 16, wherein the training data is acquired from real MRI scanners.

### Claim 18

A system for MRI reconstruction of partial Fourier k-space data using deep learning models, comprising
a memory that stores computer executable components; and
a processor that executes the computer executable components stored in the memory, wherein the computer executable components comprise:
a data acquisition component for obtaining large training data comprising complex MRI images or fully-sampled k-space data;
a training component for training artificial neural networks to predict residual complex images from incomplete k-space data; and
a reconstruction component for reconstructing high-quality MRI images from individual incomplete k-space data that are partially sampled and the residual complex images predicted by using the trained artificial neural networks.

### Claim 19

A non-transitory computer readable medium comprising executable instructions that, when executed by a processor, facilitate performance of operations, comprising:
obtaining large training data comprising complex MRI images or fully-sampled k-space data;
training artificial neural networks to predict residual complex images from incomplete k-space data; and
reconstructing high-quality MRI images from individual incomplete k-space data that are partially sampled and the residual complex images predicted by using the trained artificial neural networks.
