# US 2013/0281822 A1 · Image correction using multichannel blind deconvolution with homomorphic filtering

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

- Number: US 2013/0281822 A1
- Document ID: US20130281822A1
- Status: Published application
- Kind: A1
- Application number: 13/451,239
- Filed: 2012-04-19
- Published: 2013-10-24
- Earliest claimed: 2012-04-19
- Assignees: General Electric Company
- Inventors: Dominic Michael Graziani; Christopher Judson Hardy; Ek Tsoon Tan; Stephen Joseph Garnier
- Web page: https://patentgrep.com/patent/US20130281822A1

## CPC

- A61B 5/055: involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)

## Abstract

A method includes obtaining a plurality of magnetic resonance (MR) coil images of a subject of interest, each MR coil image being generated from one of an array of MR receiving coils; combining the plurality of coil images to generate an image estimate of the subject of interest; performing a multichannel blind deconvolution (MBD) process including: deriving coil sensitivity information for every one of the array of MR receiving coils based on the image estimate or a filtered image estimate derived from the image estimate; updating the image estimate or the filtered image estimate using the derived coil sensitivity information to generate an updated image estimate; and applying a homomorphic filter to the image estimate to derive the filtered image estimate, or to the updated image estimate to derive a filtered updated image estimate, or a combination thereof.

## Claims

### Claim 1

A method, comprising:
obtaining a plurality of magnetic resonance (MR) coil images of a subject of interest, each MR coil image being generated from data obtained by one of an array of MR receiving coils;
combining the plurality of coil images to generate an image estimate of the subject of interest;
performing a multichannel blind deconvolution (MBD) process comprising:
deriving coil sensitivity information for every one of the array of MR receiving coils based on the image estimate or a filtered image estimate derived from the image estimate; and
updating the image estimate or the filtered image estimate using the derived coil sensitivity information to generate an updated image estimate; and
applying a homomorphic filter to the image estimate to derive the filtered image estimate, or to the updated image estimate to derive a filtered updated image estimate, or a combination thereof.

### Claim 2 (depends on 1)

The method of claim 1, comprising refining the coil sensitivity information and the updated image estimate or the filtered updated image estimate by re-iterating the MBD process using the updated image estimate or the filtered updated image estimate as an input.

### Claim 3 (depends on 2)

The method of claim 2, comprising re-iterating the MBD process and the application of the homomorphic filter until a predetermined convergence criterion or criteria are met.

### Claim 4 (depends on 1)

The method of claim 1, wherein combining the plurality of coil images to generate the image estimate of the subject of interest comprises performing a sum of squares addition of the plurality of coil images.

### Claim 5 (depends on 1)

The method of claim 1, wherein the coil sensitivity information comprises a coil sensitivity map.

### Claim 6 (depends on 1)

The method of claim 1, wherein the MBD process comprises minimizing an energy function comprising:
a first term that enforces consistency within each channel, each channel comprising image data from every one of the array of receiving coils; and
a second term that enforces consistency among channels, wherein the second term comprises a matrix having pairwise comparisons between all channels.

### Claim 7 (depends on 1)

The method of claim 1, wherein obtaining the plurality of magnetic resonance (MR) coil images of the subject of interest comprises performing an MR data acquisition using the array of receiving coils, or accessing the images stored on one or more machine-readable media, or a combination thereof.

### Claim 8 (depends on 1)

The method of claim 1, wherein the homomorphic filter performs image-based intensity correction by deconvolving the original image with a low-pass filtered version of the same image.

### Claim 9 (depends on 1)

The method of claim 1, wherein obtaining a plurality of magnetic resonance (MR) coil images of a subject of interest comprises:
performing a parallel imaging acquisition in which a calibration region of k-space is fully- or oversampled and an accelerated region of k-space is subsampled such that the accelerated region comprises first data points having sampled data and second data points for which data are missing; and
generating the plurality of coil images from the calibration regions; and
wherein the derived coil sensitivity information is used to synthesize the missing data via parallel imaging reconstruction.

### Claim 10 (depends on 9)

The method of claim 9, comprising extrapolating the derived coil sensitivity information to cover an entire field of view encompassing the calibration region and the accelerated region, and using the extrapolated coil sensitivity information to synthesize the missing data.

### Claim 11 (depends on 1)

The method of claim 1, wherein obtaining a plurality of magnetic resonance (MR) coil images of a subject of interest comprises producing low resolution coil images from the data obtained by the one of the array of MR receiving coils, wherein the obtained data are at the center of k-space and do not include data outside a predetermined radius about the center of k-space, and the image estimate is updated based on sensitivity maps generated from a combined low resolution image based on the low resolution coil images.

### Claim 12 (depends on 11)

The method of claim 11, comprising interpolating the derived coil sensitivity information to the full resolution, and using the interpolated coil sensitivity information to estimate a full resolution truth image.

### Claim 13

One or more tangible, non-transitory machine-readable media comprising instructions executable by a processor to:
obtain a plurality of magnetic resonance (MR) coil images of a subject of interest, each MR coil image being generated from data obtained by one of an array of MR receiving coils;
combine the plurality of coil images to generate an image estimate of the subject of interest;
perform a multichannel blind deconvolution (MBD) process comprising:
deriving coil sensitivity information for every one of the array of MR receiving coils based on the image estimate or a filtered image estimate derived from the image estimate; and
updating the image estimate or the filtered image estimate using the derived coil sensitivity information to generate an updated image estimate; and
apply a homomorphic filter to the image estimate to derive the filtered image estimate, or to the updated image estimate to derive a filtered updated image estimate, or a combination thereof.

### Claim 14 (depends on 13)

The media of claim 13, wherein the instructions are executable by the processor to refine the coil sensitivity information and the image estimate by re-iterating the MBD process using the updated image estimate or the filtered updated image estimate as an input.

### Claim 15 (depends on 13)

The media of claim 13, wherein the instructions are executable by the processor to re-iterate the MBD process and the application of the homomorphic filter until a predetermined convergence criterion is met.

### Claim 16 (depends on 13)

The media of claim 13, wherein the instructions are executable by the processor to perform a sum of squares addition of the plurality of coil images to combine the plurality of coil images.

### Claim 17 (depends on 13)

The media of claim 13, wherein the coil sensitivity information comprises a coil sensitivity map.

### Claim 18 (depends on 13)

The media of claim 13, wherein the instructions are executable by the processor to minimize an energy function during the MBD process, and wherein the energy function comprises:
a first term that enforces consistency within each channel, each channel comprising image data from every one of the array of receiving coils; and
a second term that enforces consistency between channels, wherein the second term comprises a matrix having pairwise comparisons between all channels.

### Claim 19 (depends on 13)

The media of claim 13, wherein the instructions are executable by the processor to cause an MR imaging system to perform an MR data acquisition sequence of the subject of interest to obtain the plurality of MR coil images.

### Claim 20

A magnetic resonance (MR) imaging system, comprising:
a primary field magnet;
a plurality of gradient field coils;
a radiofrequency (RF) transmit coil;
an array of receiving coils; and
control circuitry coupled to the gradient field coils, to the RF transmit coil, and to the array of receiving coils, wherein the control circuitry is configured to:
apply control signals to the gradient, RF transmit and receiving coils to acquire a plurality of magnetic resonance (MR) coil images of the subject of interest, each MR coil image being generated from data obtained by one of the array of receiving coils;
combine the plurality of coil images to generate an image estimate of the subject of interest;
perform a multichannel blind deconvolution (MBD) process comprising:
deriving coil sensitivity information for every one of the array of MR receiving coils based on the image estimate or a filtered image estimate derived from the image estimate; and
updating the image estimate or the filtered image estimate using the derived coil sensitivity information to generate an updated image estimate; and
apply a homomorphic filter to the image estimate to derive the filtered image estimate, or to the updated image estimate to derive a filtered updated image estimate, or a combination thereof.

### Claim 21 (depends on 20)

The system of claim 20, wherein the control circuitry is configured to refine the coil sensitivity information and the image estimate by re-iterating the MBD process using the updated image estimate or the filtered updated image estimate as an input.

### Claim 22 (depends on 20)

The system of claim 20, wherein the control circuitry is configured to perform a sum of squares addition of the plurality of coil images to combine the plurality of coil images.

### Claim 23 (depends on 20)

The system of claim 20, wherein the coil sensitivity information comprises a coil sensitivity map.

### Claim 24 (depends on 20)

The system of claim 20, wherein the control circuitry is configured to minimize an energy function during the MBD process, and wherein the energy function comprises:
a first term that enforces consistency within each channel, each channel comprising image data from every one of the array of receiving coils; and
a second term that enforces consistency between channels, wherein the second term comprises a matrix having pairwise comparisons between all channels.
