# US 2023/0056685 A1 · Methods and apparatus for deep learning based image attenuation correction

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

- Number: US 2023/0056685 A1
- Document ID: US20230056685A1
- Status: Published application
- Kind: A1
- Application number: 17/759,401
- Filed: 2021-03-01
- Published: 2023-02-23
- Earliest claimed: 2020-03-04
- Assignees: Siemens Medical Solutions USA, Inc.
- Applicants: Siemens Medical Solutions USA, Inc.
- Inventors: Thomas Vahle; Matthias Fenchel
- Web page: https://patentgrep.com/patent/US20230056685A1

## CPC

- G06T 12/00: Tomographic reconstruction from projections
- G06T 12/10: Image preprocessing, e.g. calibration, positioning of sources or scatter correction
- A61B 6/037: {Emission tomography}
- G06T 2207/10104: Positron emission tomography [PET]
- G06T 2211/441: AI-based methods, deep learning or artificial neural networks
- G06T 2211/464: Dual or multimodal imaging, i.e. combining two or more imaging modalities
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)

## Abstract

Systems and methods for reconstructing medical images are disclosed. Measurement data, such as magnetic resonance (MR) data and positron emission tomography (PET) data, is received from an image scanning system. Attenuation maps are generated based on the PET data and a determined background level of radiation of the image scanning system. The background level of radiation can be caused by the radioactive decay of crystal material of the image scanning system. MR images are reconstructed based on the MR data. Further, a neural network, such as a deep learning neural network, is trained with the attenuation maps and the reconstructed MR images to determine attenuation map based on a reconstructed MR image. The trained neural network can be applied to MR data received for a patient to determine a corresponding attenuation map. A final image is generated based on PET data received for the patient and the determined attenuation map.

## Claims

### Claim 1

A computer-implemented method comprising:
receiving first positron emission tomography (PET) measurement data from an image scanning system;
determining a reference level of radiation of the image scanning system based on the first PET measurement data;
receiving magnetic resonance (MR) measurement data and second PET measurement data from the image scanning system;
generating a first attenuation map based on the first PET measurement data and the second PET measurement data;
training a neural network with the first attenuation map and the MR measurement data; and
storing the trained neural network in a memory device.

### Claim 2 (depends on 1)

The computer-implemented method of claim 1 further comprising:
receiving second MR measurement data from the image scanning system; and
applying the trained neural network to the second MR measurement data to determine a second attenuation map.

### Claim 3 (depends on 2)

The computer-implemented method of claim 2 further comprising generating an image based on the second attenuation map.

### Claim 4 (depends on 1)

The computer-implemented method of claim 1, wherein the second attenuation map is generated based on prior images computed using MR measurement data.

### Claim 5 (depends on 1)

The computer-implemented method of claim 1 wherein the first attenuation map is generated based on synthetic transmission images.

### Claim 6 (depends on 1)

The computer-implemented method of claim 1 further comprising generating the synthetic transmission images based on a detected background radiation generated by the image scanning system.

### Claim 7 (depends on 1)

The computer-implemented method of claim 1 comprising scaling the first attenuation map based on a corresponding energy window.

### Claim 8 (depends on 1)

The computer-implemented method of claim 1 wherein the neural network is a deep learning neural network.

### Claim 9

A non-transitory computer readable medium storing instructions that, when executed by at least one processor, cause the at least one processor to perform operations comprising:
receiving first positron emission tomography (PET) measurement data from an image scanning system;
determining a reference level of radiation of the image scanning system based on the first PET measurement data;
receiving magnetic resonance (MR) measurement data and second PET measurement data from the image scanning system;
generating a first attenuation map based on the first PET measurement data and the second PET measurement data;
training a neural network with the first attenuation map and the MR measurement data; and
storing the trained neural network in a memory device.

### Claim 10 (depends on 9)

The non-transitory computer readable medium of claim 9 storing instructions that, when executed by at least one processor, further cause the at least one processor to perform operations comprising:
receiving second MR measurement data from the image scanning system; and
applying the trained neural network to the second MR measurement data to determine a second attenuation map.

### Claim 11 (depends on 10)

The non-transitory computer readable medium of claim 10 storing instructions that, when executed by at least one processor, further cause the at least one processor to perform operations comprising generating an image based on the second attenuation map.

### Claim 12 (depends on 9)

The non-transitory computer readable medium of claim 9 storing instructions that, when executed by at least one processor, further cause the at least one processor to perform operations comprising generating synthetic transmission images based on a detected background radiation generated by the image scanning system, wherein the first attenuation map is generated based on the synthetic transmission images.

### Claim 13 (depends on 9)

The non-transitory computer readable medium of claim 9 wherein the second attenuation map is generated based on prior images computed using MR measurement data.

### Claim 14 (depends on 9)

The non-transitory computer readable medium of claim 9 storing instructions that, when executed by at least one processor, further cause the at least one processor to perform operations comprising scaling the first attenuation map based on a corresponding energy window.

### Claim 15

A system comprising:
a database; and
at least one processor communicatively coupled to the database and configured to:
receive first positron emission tomography (PET) measurement data from an image scanning system;
determine a reference level of radiation of the image scanning system based on the first PET measurement data;
receive magnetic resonance (MR) measurement data and second PET measurement data from the image scanning system;
generate a first attenuation map based on the first PET measurement data and the second PET measurement data;
train a neural network with the first attenuation map and the MR measurement data; and
store the trained neural network in a memory device.

### Claim 16 (depends on 15)

The system of claim 15, wherein the at least one processor is configured to:
receive second MR measurement data from the image scanning system; and
apply the trained neural network to the second MR measurement data to determine a second attenuation map.

### Claim 17 (depends on 16)

The system of claim 16, wherein the at least one processor is configured to generate an image based on the second attenuation map.

### Claim 18 (depends on 15)

The system of claim 15, wherein the at least one processor is configured to generate synthetic transmission images based on a detected background radiation generated by the image scanning system, wherein the first attenuation map is generated based on the synthetic transmission images.

### Claim 19 (depends on 15)

The system of claim 15, wherein the second attenuation map is generated based on prior images computed using MR measurement data.

### Claim 20 (depends on 15)

The system of claim 15, wherein the at least one processor is configured to scale the first attenuation map based on a corresponding energy window.
