Ct big data from simulation, emulation and transfer learning

Patent metadata

Field	Value
Publication number	US-2021035340-A1
Application number	US-201916969072-A
Country	US
Kind code	A1
Filing date	Feb 11, 2019
Priority date	Feb 12, 2018
Publication date	Feb 4, 2021
Grant date	—

Abstract

In some embodiments, a method of machine learning includes identifying, by an auto encoder network, a simulator feature based, at least in part, on a received first simulator data set and an emulator feature based, at least in part, on a received first emulator data set. The method further includes determining, by a synthesis control circuitry, a synthesized feature based, at least in part, on the simulator feature and based, at least in part, on the emulator feature; and generating, by the auto encoder network, an intermediate data set based, at least in part, on a second simulator data set and including the synthesized feature. Some embodiments of the method further include determining, by a generative artificial neural network, a synthesized data set based, at least in part, on the intermediate data set and based, at least in part, on an objective function.

First claim

1 . A method of machine learning, the method comprising:
identifying, by an auto encoder network, a simulator feature based, at least in part, on a received first simulator data set; identifying, by the auto encoder network, an emulator feature based, at least in part, on a received first emulator data set; determining, by a synthesis control circuitry, a synthesized feature based, at least in part, on the simulator feature and based, at least in part, on the emulator feature; and generating, by the auto encoder network, an intermediate data set based, at least in part, on a second simulator data set and comprising the synthesized feature.

2 . The method of claim 1 , further comprising:
determining, by a generative artificial neural network, a synthesized data set based, at least in part, on the intermediate data set and based, at least in part, on an objective function.

3 . The method of claim 1 , wherein the auto encoder network comprises an input stage, an output stage and a latent space coupled between the input stage and output stage, the simulator feature and the emulator feature extracted from the latent space and the synthesized feature provided to the latent space.

4 . The method of claim 1 , wherein the synthesized feature is determined based, at least in part, on at least one of a linear interpolation and/or an algebraic manipulation.

5 . The method of claim 1 , wherein each data set is selected from the group comprising a computed tomography (CT) sinogram and a reconstructed CT image.

6 . The method of claim 2 , wherein the objective function comprises at least one parameter related to at least one of a physical characteristic, a physiological function and/or a model of a human organ.

7 . A transfer learning apparatus comprising:
an auto encoder network configured to identify a simulator feature based, at least in part, on a received first simulator data set and an emulator feature based, at least in part, on a received first emulator data set; and a synthesis control circuitry configured to determine a synthesized feature based, at least in part, on the simulator feature and based, at least in part, on the emulator feature, the auto encoder network further configured to generate an intermediate data set based, at least in part, on a second simulator data set and comprising the synthesized feature.

8 . The apparatus of claim 7 , further comprising a generative artificial neural network configured to determine a synthesized data set based, at least in part, on the intermediate data set and based, at least in part, on an objective function.

9 . The apparatus of claim 7 , wherein the auto encoder network comprises an input stage, an output stage and a latent space coupled between the input stage and output stage, the simulator feature and the emulator feature extracted from the latent space and the synthesized feature provided to the latent space.

10 . The apparatus according to claim 7 , wherein the synthesized feature is determined based, at least in part, on at least one of a linear interpolation and/or an algebraic manipulation.

11 . The apparatus according to claim 7 , wherein each data set is selected from the group comprising a computed tomography (CT) sinogram and a reconstructed CT image.

12 . The apparatus of claim 8 , wherein the objective function comprises at least one parameter related to at least one of a physical characteristic, a physiological function and/or a model of a human organ.

13 . A machine learning system comprising:
a simulator; an emulator; and a transfer learning circuitry comprising:
an auto encoder network configured to identify a simulator feature based, at least in part, on a received first simulator data set and an emulator feature based, at least in part, on a received first emulator data set; and
a synthesis control circuitry configured to determine a synthesized feature based, at least in part, on the simulator feature and based, at least in part, on the emulator feature,
the auto encoder network further configured to generate an intermediate data set based, at least in part, on a second simulator data set and comprising the synthesized feature.

14 . The system of claim 13 , wherein the transfer learning circuitry further comprises a generative artificial neural network configured to determine a synthesized data set based, at least in part, on the intermediate data set and based, at least in part, on an objective function.

15 . The system of claim 13 , wherein the auto encoder network comprises an input stage, an output stage and a latent space coupled between the input stage and output stage, the simulator feature and the emulator feature extracted from the latent space and the synthesized feature provided to the latent space.

16 . The system according to claim 13 , wherein the synthesized feature is determined based, at least in part, on at least one of a linear interpolation and/or an algebraic manipulation.

17 . The system according to claim 13 , wherein each data set is selected from the group comprising a computed tomography (CT) sinogram and a reconstructed CT image.

18 . The system of claim 14 , wherein the objective function comprises at least one parameter related to at least one of a physical characteristic, a physiological function and/or a model of a human organ.

19 . The system according to claim 13 , wherein the simulator is a Monte Carlo simulator.

20 . A computer readable storage device having stored thereon instructions that when executed by one or more processors result in the following operations comprising the method according to claim 1 .

Assignees

• Rensselaer Polytech Inst

Inventors

• Wang Ge
• Harrison Daniel David
• Jia xun
• Mueller Klaus

Classifications

• G06T12/20

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

• G06V10/82 Primary

  using neural networks · CPC title

• G06N3/10

  Interfaces, programming languages or software development kits, e.g. for simulating neural networks · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06F18/217

  Validation; Performance evaluation; Active pattern learning techniques · CPC title