Synthetic Medical Image Generation with Conditional Generative Models

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Dynamic medical imaging can be lengthy and uncomfortable for patients due to the need to minimize movement, and it is challenging to replace poor-quality images without repeating the entire examination.

Innovation Solution

A computer-implemented method using a conditional generative model trained with medical image data sets to generate synthetic medical images based on image embeddings, allowing for the reconstruction of examination regions at different time points, reducing the need for prolonged patient immobility and image repetition.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If dynamic medical imaging is performed to capture changes in the examination region over time, then diagnostic information is improved, but examination time increases and patient comfort deteriorates

Engineering Contradiction:
Improvediagnostic informationVSAvoidexamination time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by generating synthetic intermediate time-point images from a limited set of actual captured images. The generative model is trained in advance to learn the temporal evolution patterns, enabling it to synthesize realistic intermediate frames that represent the examination region at uncaptured time points, thereby reducing the need for prolonged continuous imaging.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system creates synthetic copies of medical images at intermediate time points by using a generative model that learns from actual captured images. These synthetic images serve as realistic replicas that fill temporal gaps, allowing the system to reconstruct the dynamic examination sequence without capturing every moment, thus shortening examination duration while preserving diagnostic quality.

Inventive Principle:
Principle #26Copying

2Reliability

If dynamic medical imaging is performed to capture changes in the examination region over time, then diagnostic information is improved, but patient comfort deteriorates due to movement restriction

Engineering Contradiction:
Improvediagnostic informationVSAvoidpatient comfort
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

The system performs preliminary actions by generating synthetic intermediate time-point images from a limited set of actual captured images. The generative model is trained in advance to learn the temporal evolution patterns, enabling it to synthesize realistic intermediate frames that represent the examination region at uncaptured time points, thereby reducing the need for prolonged continuous imaging.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system creates synthetic copies of medical images at intermediate time points by using a generative model that learns from actual captured images. These synthetic images serve as realistic replicas that fill temporal gaps, allowing the system to reconstruct the dynamic examination sequence without capturing every moment, thus shortening examination duration while preserving diagnostic quality.

Inventive Principle:
Principle #26Copying

3Manufacturing precision

If traditional image replacement methods are used for poor quality images, then image quality is improved, but examination time increases due to repetition

Engineering Contradiction:
Improveimage qualityVSAvoidexamination time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The system creates synthetic copies of medical images at intermediate time points by using a generative model that learns from actual captured images. These synthetic images serve as realistic replicas that fill temporal gaps, allowing the system to reconstruct the dynamic examination sequence without capturing every moment, thus shortening examination duration while preserving diagnostic quality.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The system performs preliminary actions by generating synthetic intermediate time-point images from a limited set of actual captured images. The generative model is trained in advance to learn the temporal evolution patterns, enabling it to synthesize realistic intermediate frames that represent the examination region at uncaptured time points, thereby reducing the need for prolonged continuous imaging.

Inventive Principle:
Principle #10Preliminary action

4Measurement precision

If more time points are captured during dynamic examination, then temporal resolution is improved, but examination duration increases

Engineering Contradiction:
Improvetemporal resolutionVSAvoidexamination duration
Core Design Contradiction:
Measurement precisionVSDuration of action of stationary object

Solution Approach 1:

The system creates synthetic copies of medical images at intermediate time points by using a generative model that learns from actual captured images. These synthetic images serve as realistic replicas that fill temporal gaps, allowing the system to reconstruct the dynamic examination sequence without capturing every moment, thus shortening examination duration while preserving diagnostic quality.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The system performs preliminary actions by generating synthetic intermediate time-point images from a limited set of actual captured images. The generative model is trained in advance to learn the temporal evolution patterns, enabling it to synthesize realistic intermediate frames that represent the examination region at uncaptured time points, thereby reducing the need for prolonged continuous imaging.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4618009A1Generation of a synthetic medical image
Publication Date: 2025.09.17 BAYER AG
  • EP4618009A1 patent drawingFigure 1
  • EP4618009A1 patent drawingFigure 2
  • EP4618009A1 patent drawingFigure 3

AI summary

Systems, methods, and computer programs disclosed herein relate to training a machine learning model and using the trained machine learning model to generate a synthetic medical image of an examination region of an examination object representing the examination region at one point in time during a dynamic examination of the examination region based on one or more medical images representing the examination region at another point in time or multiple other points in time.