Synthetic Medical Image Artifact Detection by Confidence Dispersion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Medical images generated by machine learning models may contain errors, leading to uncertainties in diagnosis and therapy, as physicians struggle to distinguish between real features and artifacts.
Innovation Solution
A method to generate synthetic images using a generative model, applying image modifications to assess the trustworthiness by determining a confidence value based on the dispersion of color values of corresponding image elements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If machine learning models are used to generate synthetic medical images, then productivity of medical examinations is improved, but reliability of diagnostic results deteriorates due to artifacts
Solution Approach 1:
The system generates multiple synthetic images from modified input images and uses dispersion analysis to provide feedback on the reliability of each generated image. This feedback mechanism allows the system to identify and flag artifacts, thereby maintaining diagnostic accuracy while preserving the productivity benefits of synthetic image generation.
Solution Approach 2:
The patent introduces an intermediary verification step between image generation and diagnostic use. By generating multiple variants and analyzing their dispersion, the system creates an intermediate reliability assessment that mediates between the productive synthetic generation process and the reliable diagnostic decision-making process.
2Measurement precision
If multiple synthetic images are generated to assess reliability, then measurement precision of trustworthiness is improved, but device complexity increases
Solution Approach 1:
The system segments the reliability assessment into discrete, manageable components by analyzing each image element's color value dispersion independently. This segmentation allows for precise measurement of trustworthiness at the pixel level while keeping the overall processing complexity manageable through modular computation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure relates to the technical field of generating synthetic medical images. The subject matter of the present disclosure includes a method, a computer system, and a computer-readable storage medium comprising a computer program for detecting artifacts in synthetic medical images.