Table 1 Representative patent literature for image reconstruction using AI technology

WO2017223560A1

Tomographic image reconstruction via ML

A ML method is proposed to improve the quality of tomographic images by applying ML models on raw data, processed data, or an intermediate image to reduce noise and artifacts.

RPI

06–24-16

WO2018126396A1

DL based estimation of data for use in tomographic reconstruction

Trained neural network is used to estimate various types of missing projection data.

GE

01–05-17

US2018197317A1

DL based acceleration for iterative tomographic reconstruction

A DL technique is used to accelerate iterative reconstruction of images.

GE

01–06-17

US2019102916A1

Systems and methods for DL-based image reconstruction

A method includes acquiring a set of imaging projections data, identifying a voxel to be reconstructed, receiving a trained regression model, and reconstructing the voxel.

GE

09–29-17

US2021074033A1

DL-based data rescue in emission tomography medical imaging

An emission image is generated from poor quality emission data. A machine-learned model is used to recover information related to the data.

Siemens

09–09-19

US2017372193A1

Image correction using a deep generative machine-learning model

A deep-learnt generative model is used as a regularizer in an inverse solution with a physics model of the degradation behavior of the imaging system. The generative model is trained from good images, so difficulty gathering problem-specific training data may be avoided or reduced.

Siemens

06–23-16

US2020311490A1

Apparatus and method for sinogram restoration in CT using adaptive filtering with DL

A method is proposed to train a DL network to optimize the convolution kernel of an adaptive filter that is applied in the data domain. The input data can be sinograms generated by a low-dose CT scan, and the target data generated by a high-dose CT scan.

Canon

04–01-19

US2021012541A1

Apparatus and method using DL to improve analytical tomographic image reconstruction

A method is proposed to improve the image quality of images generated by analytical reconstruction of a CT image. This improved image quality results from a DL network that is used to filter a sinogram before back projection but after the sinogram has been filtered using a ramp filter or other reconstruction kernel.

Canon

07–11-19

US2021192809A1

Tomographic image reconstruction using AI engines

A method includes obtaining two-dimensional (2D) projection data and processing the 2D projection data using the AI engine. AI engine may involve: generating 2D feature data by processing the 2D projection data using the multiple first processing layers, reconstructing first three-dimensional (3D) feature volume data from the 2D feature data using the back-projection module; and generating second 3D feature volume data by processing the first 3D feature volume data using the multiple second processing layers.

Varian Medical System

12–20-19

EP3367329A1

Denoising medical images by learning sparse image representations with a deep unfolding approach

Method is to learn sparse image representations with deep unfolding and applying the machine learnt network medical image denoising.

Siemens

02–22-17

WO2019060843A1

Image reconstruction using ML regularizers

A method for reconstructing an image of a target object using an iterative reconstruction technique can use a ML model as a regularization filter.

Nview Medical Inc

09–22-17

US2021118204A1

Method for reconstructing incomplete data of X-ray absorption contrast computed tomography based on DL

A DL-based method is proposed using a filtered back projection (FBP) algorithm to obtain an initial reconstructed image; forward projecting the initial reconstructed image to obtain artifact-contaminated complete projection sequences; using a DL technique to process the artifact-contaminated projection sequences to obtain artifact-free projection sequences; using the FBP algorithm to reconstruct the artifact-free projection sequences to obtain a final reconstructed image.

Beihang of University

10–18-19

US2021118200A1

Systems and methods for training ML algorithms for inverse problems without fully sampled reference data

A self-supervised training of ML algorithm is proposed for reconstruction in inverse problems. A physics-based ML reconstruction can be trained without requiring fully-sampled training data.

University of Minnesota

10–21-19

US2018018757A1

Transforming projection data in tomography by means of ML

A method is proposed to use a ML model to outputs high quality projection data from low quality data.

Suzuki Kenji

07–13-16

WO2018236748A1

DL-assisted image reconstruction for tomographic imaging

An iterative image reconstruction method produces a plurality of intermediate images and to produce the image of the subject. One selected intermediate image from the plurality of intermediate images using a quasi- projection operator. The quasi-projection operator uses a deep-learning model configured to map the at least one selected intermediate image to at least one regularized intermediate image.

University of Washington

06–19-17

EP2890300B1

Supervised ML technique for reduction of radiation dose in computed tomography imaging

A technique is proposed for converting low-dose CT images to higher quality, lower noise images using ML.

University Chicago

08–31-12

WO2019074879A1

Image generation using ML

A ML model is used to convert an input image generated from a computationally efficient algorithm to a final high quality image.

GE

10–11-17

WO2018187020A1

Tomographic reconstruction based on DL

Tomographic images are used as an input to a neural network. More layers of the neural network are used as wavelet filter banks.

GE

04–05-17

US2019104940A1

Apparatus and method for medical image reconstruction using DL for CT image noise and artifacts reduction

A method is proposed to reduce noise and artifacts in reconstructed medical images using a DL network.

Canon

10–06-17