# US 12,223,568 B2 · Systems and methods for generating and estimating unknown and unacquired ultrasound data

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

- Number: US 12,223,568 B2
- Document ID: US12223568B2
- Status: Granted
- Kind: B2
- Application number: 17/257,936
- Filed: 2019-07-08
- Published: 2021-09-02
- Granted: 2025-02-11
- Earliest claimed: 2018-07-06
- Assignees: Mayo Foundation for Medical Education and Research
- Applicants: Mayo Foundation for Medical Education and Research
- Inventors: Bae-Hyung Kim; Azra Alizad; Mostafa Fatemi; Viksit Kumar
- Web page: https://patentgrep.com/patent/US12223568B2

## CPC

- A61B 8/5207: {involving processing of raw data to produce diagnostic data, e.g. for generating an image}
- G06T 12/30: Image post-processing, e.g. metal artefact correction
- A61B 8/4245: {involving determining the position of the probe, e.g. with respect to an external reference frame or to the patient}
- A61B 8/5246: {combining images from the same or different imaging techniques, e.g. color Doppler and B-mode}
- A61B 8/5269: {involving detection or reduction of artifacts}
- G01S 7/52034: {Data rate converters}
- G01S 7/52047: {for elimination of side lobes or of grating lobes; for increasing resolving power}
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/09: Supervised learning
- G06N 20/00: Machine learning
- G01S 15/8915: {using a transducer array}
- G01S 15/8995: {Combining images from different aspect angles, e.g. spatial compounding}
- G01S 15/8997: {using synthetic aperture techniques}
- G06N 3/08: Learning methods
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G01S: RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Systems and methods for reconstructing, estimating, or otherwise generating unacquired, undetected, unreconstructed, or otherwise unknown ultrasound data using machine learning algorithms are provided. Thus, the systems and methods described in the present disclosure provide for generating unacquired, undetected, unreconstructed, or otherwise unknown data that are not actually and physically acquired with an ultrasound transducer and/or front-end receiver of an ultrasound system.

## Claims

### Claim 1

A method for reconstructing an image from data acquired with an ultrasound system, comprising:
(a) accessing with a computer system, ultrasound data acquired from a subject with an ultrasound system, wherein the ultrasound data correspond to one or more data channels of the ultrasound system, wherein each data channel of the ultrasound system corresponds to a transducer element of a transducer array of the ultrasound system;
(b) generating additional ultrasound data with the computer system by inputting the ultrasound data to a trained machine learning algorithm that is trained to generate additional ultrasound data based on the acquired ultrasound data, generating output as the additional ultrasound data, wherein the additional ultrasound data correspond to data channels of the ultrasound system that were not sampled when acquiring the ultrasound data;
(c) reconstructing with the computer system, an image from the ultrasound data and the additional ultrasound data, wherein the image depicts the subject from which the ultrasound data were acquired.

### Claim 2 (depends on 1)

The method as recited in claim 1, wherein the additional ultrasound data comprise data corresponding to unacquired channel data.

### Claim 3 (depends on 1)

The method as recited in claim 1, wherein the additional data comprise data corresponding to unacquired steering angle data.

### Claim 4 (depends on 1)

The method as recited in claim 1, wherein the additional data comprise data corresponding to pseudo channel data associated with an arbitrary position on the transducer array.

### Claim 5 (depends on 4)

The method as recited in claim 4, wherein the arbitrary position corresponds to a virtual channel between two physical channels.

### Claim 6 (depends on 1)

The method as recited in claim 1, wherein the trained machine learning algorithm comprises a deep learning model.

### Claim 7 (depends on 6)

The method as recited in claim 6, wherein the deep learning model comprises a plurality of SR blocks.

### Claim 8 (depends on 7)

The method as recited in claim 7, wherein each SR block comprises at least one convolutional layer and at least one activation function layer.

### Claim 9 (depends on 8)

The method as recited in claim 8, wherein the at least one activation function layer comprises a leaky rectified linear unit (ReLU).

### Claim 10 (depends on 8)

The method as recited in claim 8, wherein each SR block further comprises at least one batch normalization layer.

### Claim 11 (depends on 7)

The method as recited in claim 7, wherein the deep learning model further comprises a plurality of Trans layers.

### Claim 12 (depends on 11)

The method as recited in claim 11, wherein each Trans layer receives an input from one of the plurality of SR blocks.

### Claim 13 (depends on 7)

The method as recited in claim 7, wherein each SR block includes a skip connection that concatenates an input to the SR block with an output of the SR block.

### Claim 14 (depends on 7)

The method as recited in claim 7, wherein the deep learning model includes a skip connection that concatenates the acquired ultrasound data with an output of the plurality of SR blocks.

### Claim 15 (depends on 7)

The method as recited in claim 7, wherein each SR block receives input data, captures local features from the input data, merges the local features with global features in the input data, and generates output as merged local features and global features.