# US 2023/0410313 A1 · Image data processing to increase follow-up analysis fidelity

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

- Number: US 2023/0410313 A1
- Document ID: US20230410313A1
- Status: Published application
- Kind: A1
- Application number: 18/458,416
- Filed: 2023-08-30
- Published: 2023-12-21
- Earliest claimed: 2020-08-24
- Assignees: GE Precision Healthcare LLC
- Applicants: GE Precision Healthcare LLC
- Inventors: Raz Carmi
- Web page: https://patentgrep.com/patent/US20230410313A1

## CPC

- G06T 7/0016: {involving temporal comparison}
- G06F 18/2148: {characterised by the process organisation or structure, e.g. boosting cascade}
- G06N 3/045: Combinations of networks
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/088: Non-supervised learning, e.g. competitive learning
- G06N 3/0895: Weakly supervised learning, e.g. semi-supervised or self-supervised learning
- G06V 10/70: using pattern recognition or machine learning (optical pattern recognition or electronic computations therefor G06V10/88)
- G16H 30/20: for handling medical images, e.g. DICOM, HL7 or PACS
- G16H 30/40: for processing medical images, e.g. editing
- G06T 2207/10104: Positron emission tomography [PET]
- G06T 2207/20008: Globally adaptive
- G06T 2207/20081: Training; Learning
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/30096: Tumor; Lesion
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA

## Abstract

Techniques are provided for improving image data quality, such as in functional imaging follow-up studies, using reconstruction, post-processing, and/or deep-learning enhancement approaches in a way that automatically improves analysis fidelity, such as lesion tracking fidelity. The disclosed approaches may be useful in improving the performance of automatic analysis methods as well as in facilitating reviews performed by clinician.

## Claims

### Claim 1

An image processing system, the image processing system comprising:
one or more memory structures storing processor-executable code; and
one or more processors in communication with the one or more memory structures, wherein the processor-executable code, when executed by the one or more processors, cause the one or more processors to perform acts comprising:
training a deep learning network model for image enhancement by iteratively performing steps comprising:
receiving a first input image volume and a second input image volume;
providing the first input image volume and the second input image volume to the deep learning network model, wherein the deep learning network model is configured to respectively output a first output image volume and a second output image volume;
comparing the first output image volume and the second output image volume based on a fidelity metric comprising a measure of structural similarity, wherein a loss function for the deep learning network model is based on the fidelity metric and training the deep learning network model is based on reduction or minimization of the loss function.

### Claim 2 (depends on 1)

The image processing system of claim 1, wherein the first input image volume or sub-volume comprises a first version of a synthetic three-dimensional (3D) structure and the second input image volume or sub-volume together comprises a second version of the synthetic 3D structure, wherein the first version of the synthetic 3D structure and the second version of the synthetic 3D structure comprise a shared base shape but differ based on different random noise or other random artifacts being applied to each of the first version and second version.

### Claim 3 (depends on 1)

The image processing system of claim 1, wherein the processor-executable code, when executed by the one or more processors, cause the one or more processors to perform further acts comprising:
providing an input image volume to the deep learning network model to generate an enhanced image volume; and
providing the enhanced image volume for review as part of a follow-up analysis.

### Claim 4 (depends on 1)

The image processing system of claim 1, wherein the deep learning network model is configured as a Siamese network.

### Claim 5 (depends on 1)

The image processing system of claim 1, wherein training the deep learning network model does not utilize ground-truth training images.
