Techniques for automatically characterizing liver tissue of a patient

Patent metadata

Field	Value
Publication number	US-11861827-B2
Application number	US-202117168545-A
Country	US
Kind code	B2
Filing date	Feb 5, 2021
Priority date	Feb 6, 2020
Publication date	Jan 2, 2024
Grant date	Jan 2, 2024

Abstract

The disclosure relates to techniques for automatically characterizing liver tissue of a patient, comprising receiving morphological magnetic resonance image data set and at least one magnetic resonance parameter map of an imaging region comprising at least partially the liver of the patient, each acquired by a magnetic resonance imaging device, via a first interface. The techniques further include applying a trained function comprising a neural network to input data comprising at least the image data set and the parameter map. At least one tissue score describing the liver tissue is generated as output data, which is provided using a second interface.

First claim

What is claimed is:

1. A computer-implemented method for automatically characterizing liver tissue of a patient, comprising:
receiving, via a first interface, a morphological magnetic resonance image data set and a magnetic resonance parameter map,
wherein the morphological magnetic resonance image data set and the magnetic resonance parameter map are each acquired via a magnetic resonance imaging device;
applying, via one or more processors, a trained function comprising a neural network to input data, the input data comprising the morphological magnetic resonance image data set and the magnetic resonance parameter map; and
providing, via a second interface, generated output data comprising a tissue score describing the liver tissue,
wherein the input data further comprises a magnetic resonance elastography data set of an imaging region comprising at least a portion of the liver tissue of the patient.

2. The method according to claim 1 , wherein the tissue score comprises one or more of an nonalcoholic fatty liver disease (NAFLD) activity score, a steatosis score, a lobular inflammation score, a hepatocyte ballooning score, a fibrosis score, and/or a fibrosis stage.

3. The method according to claim 1 , wherein the morphological magnetic resonance image data set comprises an image data set obtained via a fat-water separation technique.

4. The method according to claim 3 , wherein the fat-water separation technique comprises a Dixon technique.

5. The method according to claim 1 , wherein a parameter of the magnetic resonance parameter map includes one or more of relaxation times, a reciprocal of relaxation times, an extracellular fluid measure, and/or a fat fraction.

6. The method according to claim 5 , wherein the reciprocal of relaxation times comprises R2*.

7. The method according to claim 1 , wherein the input data further comprises a magnetic resonance diffusion.

8. The method according to claim 1 , wherein the input data further comprises scalar and/or vector information related to the patient and/or related to the liver tissue of the patient.

9. The method according to claim 8 , wherein the scalar and/or vector information includes one or more of demographic information, medical history information, and/or laboratory results.

10. The method according to claim 8 , wherein the scalar and/or vector information is received from an electronic health record of the patient.

11. The method according to claim 1 , wherein the neural network comprises:
a convolutional layer in at least one convolutional partial neural network configured to extract features from the morphological magnetic resonance image data set and/or the magnetic resonance parameter map; and
a fully connected layer in a dense partial neural network configured to derive the output data using the extracted features.

12. The method according to claim 11 , wherein the one or more processors are further configured to, when the input data comprises scalar and/or vector information, add the scalar and/or vector information, or intermediate data derived therefrom, to a feature vector generated by the at least one convolutional partial neural network.

13. The method according to claim 12 , wherein the intermediate data is generated by an additional dense partial neural network having at least one fully connected layer.

14. The method according to claim 1 , wherein the output data further comprises predictive outcome information including risk scores for events related to liver tissue.

15. The method according to claim 1 , wherein the trained function further comprises an uncertainty estimation subfunction that is configured to determine at least one uncertainty information regarding the output data, and
wherein the second interface is further configured to provide the output data including the at least one uncertainty information.

16. The method according to claim 1 , wherein the morphological magnetic resonance image data set and the magnetic resonance parameter map are from among a plurality of morphological magnetic resonance image data sets and a plurality of magnetic resonance parameter maps, respectively, and
wherein the one or more processors are configured to perform a preprocessing step comprising:
registering each one of the plurality of morphological magnetic resonance image data sets and each one of the plurality of magnetic resonance parameter maps to one other; and/or
segmenting a region of interest to be analyzed by the trained function in each one of the plurality of morphological magnetic resonance image data sets and each one of the plurality of magnetic resonance parameter maps.

17. A characterization system for automatically characterizing liver tissue of a patient, comprising:
a first interface configured to receive a morphological magnetic resonance image data set and a magnetic resonance parameter map,
wherein the morphological magnetic resonance image data set and the magnetic resonance parameter map are each associated with an imaging region comprising at least a portion of the liver tissue of the patient and are acquired via a magnetic resonance imaging device;
one or more processors configured to apply a trained function comprising a neural network to input data, the input data comprising the morphological magnetic resonance image data set and the magnetic resonance parameter map; and
a second interface configured to provide generated output data comprising a tissue score describing the liver tissue,
wherein the input data further comprises a magnetic resonance elastography data set of an imaging region comprising at least a portion of the liver tissue of the patient.

18. A non-transitory computer-readable medium comprising instructions which, when executed by one or more processors of a characterization system, cause the characterization system to:
receive, via a first interface, a morphological magnetic resonance image data set and a magnetic resonance parameter map,
wherein the morphological magnetic resonance image data set and the magnetic resonance parameter map are each associated with an imaging region comprising at least a portion of a liver tissue of a patient and are acquired via a magnetic resonance imaging device;
apply a trained function comprising a neural network to input data, the input data comprising the morphological magnetic resonance image data set and the magnetic resonance parameter map; and
provide, via a second interface, generated output data comprising a tissue score describing the liver tissue of the patient,
wherein the input data further comprises a magnetic resonance elastography data set of an imaging region comprising at least a portion of the fiver tissue of the patient.

19. The method according to claim 1 , wherein the magnetic resonance elastography data set provides information related to a stiffness of the liver tissue and is acquired using magnetic resonance elastography (MRE).

20. The method according to claim 7 , wherein magnetic resonance diffusion-weighted data set comprises an apparent diffusion coefficient (ADC) map.

Assignees

• Siemens Healthcare Gmbh

Inventors

• Kannengiesser Stephan
• Kiefer Berthold
• Mansi Tommaso
• Nickel Marcel Dominik
• Pheiffer Thomas

Classifications

• A61B5/055

  involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06T7/0012 Primary

  Biomedical image inspection · CPC title

• A61B5/4244

  liver · CPC title