Tomographic image analysis using artificial intelligence (AI) engines

Patent metadata

Field	Value
Publication number	US-11776172-B2
Application number	US-202217721348-A
Country	US
Kind code	B2
Filing date	Apr 14, 2022
Priority date	Dec 20, 2019
Publication date	Oct 3, 2023
Grant date	Oct 3, 2023

Abstract

Example methods and systems for tomographic data analysis are provided. One example method may comprise: obtaining first three-dimensional (3D) feature volume data and processing the first 3D feature volume data using an AI engine that includes multiple first processing layers, an interposing forward-projection module and multiple second processing layers. Example processing using the AI engine may involve: generating second 3D feature volume data by processing the first 3D feature volume data using the multiple first processing layers, transforming the second 3D volume data into 2D feature data using the forward-projection module and generating analysis output data by processing the 2D feature data using the multiple second processing layers.

First claim

We claim:

1. A method for a computer system to perform tomographic data analysis, wherein the method comprises:
obtaining two-dimensional (2D) projection data associated with a target object and acquired using an imaging system; and
generating analysis output data by processing the 2D projection data using an integrated AI engine that includes (a) a first AI engine and (b) a second AI engine by performing the following:
based on the 2D projection data, generating first 2D feature data using multiple first processing layers of the first AI engine that operate in a 2D projection space;
based on the first 2D feature data, generating first 3D feature volume data using multiple second processing layers of the first AI engine that operate in a 3D volume space;
based on the first 3D feature volume data, generating second 3D feature volume data using multiple third processing layers of the second AI engine that operate in the 3D volume space; and
based on the second 3D feature volume data, generating the analysis output data using multiple fourth processing layers of the second AI engine that operate in the 2D projection space.

2. The method of claim 1 , wherein generating the first 3D feature volume data comprises:
reconstructing third 3D feature volume data from the first 2D feature data using a back-projection module of the first AI engine; and
generating the first 3D feature volume data by processing the third 3D feature volume data using the multiple second processing layers of the first AI engine that are trained to perform encoding on the third 3D feature volume data.

3. The method of claim 2 , wherein generating the first 2D feature data comprises:
generating the first 2D feature data by processing the 2D projection data using the multiple first processing layers that are trained to perform one of the following: pre-processing on the 2D projection data; and apply one or more convolution filters on the 2D projection data.

4. The method of claim 1 , wherein generating the analysis output data comprises:
transforming the second 3D feature volume data into second 2D feature data using a forward-projection module of the first AI engine; and
generating the analysis output data by processing the second 2D feature data using the multiple fourth processing layers of the second AI engine.

5. The method of claim 1 , wherein generating the second 3D feature volume data comprises:
generating the second 3D feature volume data using multiple third processing layers of the second AI engine that are trained to perform encoding on the first 3D feature volume data.

6. The method of claim 1 , wherein the method comprises at least one of the following:
training the first AI engine by training the multiple first processing layers and the multiple second processing layers, with a forward-projection module interposed in between, to (a) transform training 2D projection data into training 3D feature volume data, and (b) learn first weight data associated with the multiple first processing layers and second weight data associated with the multiple second processing layers; and
training the second AI engine by training the multiple third processing layers and the multiple fourth processing layers, with a forward-projection module interposed in between, to (a) transform training 3D feature volume data into training analysis output data, and (b) learn third weight data associated with the multiple third processing layers and fourth weight data associated with the multiple fourth processing layers.

7. The method of claim 1 , wherein the method comprises the following:
training the first AI engine and the second AI engine of the integrated AI engine together to (a) transform training 2D projection data into training analysis output data, and (b) learn first weight data associated with the multiple first processing layers, second weight data associated with the multiple second processing layers, third weight data associated with the multiple third processing layers and fourth weight data associated with the multiple fourth processing layers.

8. A non-transitory computer-readable storage medium that includes a set of instructions which, in response to execution by a processor of a computer system, cause the processor to perform a method of tomographic data analysis, wherein the method comprises:
obtaining two-dimensional (2D) projection data associated with a target object and acquired using an imaging system; and
generating analysis output data by processing the 2D projection data using an integrated AI engine that includes (a) a first AI engine and (b) a second AI engine by performing the following:
based on the 2D projection data, generating first 2D feature data using multiple first processing layers of the first AI engine that operate in a 2D projection space;
based on the first 2D feature data, generating first 3D feature volume data using multiple second processing layers of the first AI engine that operate in a 3D volume space;
based on the first 3D feature volume data, generating second 3D feature volume data using multiple third processing layers of the second AI engine that operate in the 3D volume space; and
based on the second 3D feature volume data, generating the analysis output data using multiple fourth processing layers of the second AI engine that operate in the 2D projection space.

9. The non-transitory computer-readable storage medium of claim 8 , wherein generating the first 3D feature volume data comprises:
reconstructing third 3D feature volume data from the first 2D feature data using a back-projection module of the first AI engine; and
generating the first 3D feature volume data by processing the third 3D feature volume data using the multiple second processing layers of the first AI engine that are trained to perform encoding on the third 3D feature volume data.

10. The non-transitory computer-readable storage medium of claim 9 , generating the first 2D feature data comprises:
generating the first 2D feature data by processing the 2D projection data using the multiple first processing layers that are trained to perform one of the following: pre-processing on the 2D projection data; and apply one or more convolution filters on the 2D projection data.

11. The non-transitory computer-readable storage medium of claim 8 , wherein generating the analysis output data comprises:
transforming the second 3D feature volume data into second 2D feature data using a forward-projection module of the first AI engine; and
generating the analysis output data by processing the second 2D feature data using the multiple fourth processing layers of the second AI engine.

12. The non-transitory computer-readable storage medium of claim 8 , wherein generating the second 3D feature volume data comprises:
generating the second 3D feature volume data using multiple third processing layers of the second AI engine that are trained to perform encoding on the first 3D feature volume data.

13. The non-transitory computer-readable storage medium of claim 8 , wherein the method comprises at least one of the following:
training the first AI engine by training the multiple first processing layers and the multiple second processing layers, with a forward-projection module interposed in between, to (a) transform training 2D projection data into training 3D feature volume data, and (b) learn first weight data associated with the multiple first processing layers and second weight data associated with the multiple second processing layers; and
training the second AI engine by training the multiple third processing layers and the multiple fourth processing layers, wit

Assignees

• Siemens Healthineers Int Ag

Inventors

• Paysan Pascal
• Haas Benjamin M
• Nord Janne
• Perttu Sami Petri
• Seghers Dieter
• Pyyry Joakim

Classifications

• G06T12/10

  Image preprocessing, e.g. calibration, positioning of sources or scatter correction · CPC title

• G06T12/20 Primary

  Inverse problem, i.e. transformations from projection space into object space · CPC title

• G06T12/30 Primary

  Image post-processing, e.g. metal artefact correction · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title