# US 2021/0279847 A1 · Systems and methods for adaptive blending in computed tomography imaging

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

- Number: US 2021/0279847 A1
- Document ID: US20210279847A1
- Status: Published application
- Kind: A1
- Application number: 16/807,586
- Filed: 2020-03-03
- Published: 2021-09-09
- Earliest claimed: 2020-03-03
- Assignees: GE Precision Healthcare LLC
- Applicants: GE Precision Healthcare LLC
- Inventors: Brian Edward Nett; Jed Douglas Pack; Zhye Yin; Jie Tang
- Web page: https://patentgrep.com/patent/US20210279847A1

## CPC

- G06T 5/50: using two or more images, e.g. averaging or subtraction
- A61B 6/027: {characterised by the use of a particular data acquisition trajectory, e.g. helical or spiral}
- A61B 6/5205: {involving processing of raw data to produce diagnostic data}
- A61B 6/5229: {combining image data of a patient, e.g. combining a functional image with an anatomical image}
- A61B 6/5258: {involving detection or reduction of artifacts or noise}
- G01N 23/046: using tomography, e.g. computed tomography [CT]
- G06T 5/20: using local operators
- G06T 5/60: using machine learning, e.g. neural networks
- G06T 5/70: Denoising; Smoothing
- G06T 12/20: Inverse problem, i.e. transformations from projection space into object space
- G06T 12/30: Image post-processing, e.g. metal artefact correction
- A61B 6/032: {Transmission computed tomography [CT]}
- G01N 2223/401: image processing
- G06T 2207/10081: Computed x-ray tomography [CT]
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/20221: Image fusion; Image merging
- G06T 2207/30004: Biomedical image processing
- G06T 2210/41: Medical
- G06T 2211/421: Filtered back projection [FBP]
- G06T 2211/424: Iterative
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)
- G01N: INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES (measuring or testing apparatus or processes other than immunoassay, involving enzymes or microorganisms C12M, C12Q)

## Abstract

Systems and methods are provided for computed tomography (CT) imaging. In one embodiment, a method comprises adaptively blending at least two input image volumes with different spatially-variant noise characteristics to generate an output image volume with uniform noise throughout the output image volume. In this way, images may be reconstructed from projection data with data redundancy without introducing image artifacts from stitching images or variance in image noise due to the data redundancy.

## Claims

### Claim 1

A method, comprising:
adaptively blending at least two input image volumes with different spatially-variant noise characteristics to generate an output image volume with uniform noise throughout the output image volume.

### Claim 2 (depends on 1)

The method of claim 1, further comprising reconstructing a first image volume and a second image volume from a first set of projection data, wherein image noise of the second image volume is reduced relative to image noise of the first image volume, and wherein the at least two input image volumes comprises the first image volume and the second image volume.

### Claim 3 (depends on 2)

The method of claim 2, further comprising reconstructing a third image volume and a fourth image volume from a second set of projection data, wherein image noise of the fourth image volume is reduced relative to image noise of the third image volume, and wherein the at least two input image volumes further comprise the third image volume and the fourth image volume.

### Claim 4 (depends on 3)

The method of claim 3, wherein the first image volume and the second image volume at least partially overlap the third image volume and the fourth image volume at an overlap region, and wherein image noise in the overlap region in the output image volume is equal to image noise away from the overlap region in the output image volume.

### Claim 5 (depends on 2)

The method of claim 2, further comprising reconstructing the first image volume via filtered backprojection, and processing the second image volume with one or more noise-reduction methods to reduce the image noise of the second image volume.

### Claim 6 (depends on 2)

The method of claim 2, further comprising reconstructing the first image volume via one of filtered backprojection with a first convolution kernel, an iterative reconstruction algorithm, and a convolutional neural network trained to perform image reconstruction, and reconstructing the second image volume via one of the iterative reconstruction algorithm, the convolutional neural network, or filtered backprojection with a second convolution kernel.

### Claim 7 (depends on 1)

The method of claim 1, wherein adaptively blending the at least two input image volumes comprises calculating weights for blending the at least two input image volumes, and combining the at least two input image volumes weighted by the weights to generate the output image volume.

### Claim 8 (depends on 7)

The method of claim 7, further comprising calculating the weights for blending the at least two input image volumes based on acquisition geometry.

### Claim 9 (depends on 8)

The method of claim 8, wherein the at least two input image volumes are reconstructed from projection data acquired during a helical scan of a subject, and wherein calculating the weights based on the acquisition geometry comprises calculating the weights based on a starting location for a table upon which the subject is positioned, a starting view angle for the helical scan, an ending view angle for the helical scan, and a helical pitch of the helical scan.

### Claim 10 (depends on 8)

The method of claim 8, wherein the at least two input image volumes are reconstructed from projection data acquired during an axial scan of a subject, and wherein calculating the weights based on the acquisition geometry comprises calculating the weights based on a starting view angle of the axial scan and an ending view angle of the axial scan.

### Claim 11 (depends on 10)

The method of claim 10, wherein calculating the weights based on the acquisition geometry further comprises calculating the weights based on an overlap region between the at least two input image volumes.

### Claim 12 (depends on 8)

The method of claim 8, further comprising calculating noise-neutral blending weights for blending the at least two input image volumes, wherein adaptively blending the at least two input image volumes comprises blending the at least two input image volumes according to the noise-neutral blending weights to obtain at least two noise-adjusted image volumes, and blending the at least two noise-adjusted image volumes based on the weights to obtain the output image volume.

### Claim 13

A method, comprising:
reconstructing a first image volume from a first set of projection data acquired during a scan of a subject;
reconstructing a second image volume from a second set of projection data acquired during the scan;
denoising the first image volume to generate a first denoised image volume;
denoising the second image volume to generate a second denoised image volume;
blending the first image volume and the first denoised image volume based on noise-neutral blending weights to generate a first noise-adjusted image volume;
blending the second image volume and the second denoised image volume based on the noise-neutral blending weights to generate a second noise-adjusted image volume; and
blending the first noise-adjusted image volume and the second noise-adjusted image volume based on locational blending weights to generate a noise-neutral blended image.

### Claim 14 (depends on 13)

The method of claim 13, further comprising calculating the locational blending weights based on an overlap between the first image volume and the second image volume.

### Claim 15 (depends on 13)

The method of claim 13, further comprising calculating the noise-neutral blending weights based on the locational blending weights, a prescribed blending ratio, and a tuning parameter related to noise levels in the first image volume and the first denoised image volume.

### Claim 16 (depends on 13)

The method of claim 13, wherein denoising the first image volume and the second image volume comprises filtering the first image volume and the second image volume with non-linear image filters.

### Claim 17

A system, comprising:
an x-ray source configured to generate a beam of x-rays towards a subject;
a detector array comprising a plurality of detector elements configured to detect the beam of x-rays attenuated by the subject; and
a computing device communicatively coupled to the x-ray source and the detector array, the computing device configured with instructions in non-transitory memory that when executed cause the computing device to:
adaptively blend at least two input image volumes, reconstructed from projection data acquired via the detector array, with different spatially-variant noise characteristics to generate an output image volume with uniform noise throughout the output image volume.

### Claim 18 (depends on 17)

The system of claim 17, wherein the computing device is further configured with instructions in the non-transitory memory that when executed cause the computing device to: reconstruct a first image volume and a second image volume from the projection data, wherein image noise of the second image volume is reduced relative to image noise of the first image volume, and wherein the at least two input image volumes comprises the first image volume and the second image volume.

### Claim 19 (depends on 18)

The system of claim 18, wherein the computing device is further configured with instructions in the non-transitory memory that when executed cause the computing device to: reconstruct the first image volume via filtered backprojection, and process the second image volume with one or more noise-reduction methods to reduce the image noise of the second image volume.

### Claim 20 (depends on 17)

The system of claim 17, wherein the computing device is further configured with instructions in the non-transitory memory that when executed cause the computing device to: reconstruct the first image volume via filtered backprojection, and reconstruct the second image volume via one of an iterative reconstruction algorithm, a convolutional neural network trained to reconstruct the second image volume with reduced noise, or filtered backprojection with a different convolution kernel.