Multi-Scale GAN for Medical Image Artifact Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Medical imaging technologies face challenges in removing global and local artifacts from diagnostic images, which can degrade image quality, lead to misinterpretation, and increase patient exposure and stress due to the need for re-imaging, and current generative adversarial networks (GANs) are limited in generating high-resolution images.
Innovation Solution
A multi-scale deep learning approach utilizing multiple GANs to down-sample images, generate high-resolution patches, and integrate them seamlessly to produce high-resolution images with reduced global and local artifacts, improving image quality and efficiency for diagnostic purposes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current generative adversarial networks are used to remove artifacts, then artifact removal is achieved, but image resolution is degraded
Solution Approach 1:
The patent segments the image processing task into multiple stages: first removing global artifacts using a global artifact removal network, then removing local artifacts using a local artifact removal network, and finally reconstructing high-resolution details using a high-resolution network. This multi-stage segmentation allows each network to specialize in specific artifact types while preserving overall image quality and resolution.
Solution Approach 2:
The patent introduces a new dimension of processing by implementing a multi-scale architecture that operates at different resolution levels. The system processes images at both low resolution (for global artifact removal) and high resolution (for local artifact removal and detail reconstruction), effectively adding a resolution dimension to the artifact removal process.
2Reliability
If multiple imaging procedures are performed to ensure diagnostic accuracy, then diagnostic reliability is improved, but patient exposure and stress increase
Solution Approach 1:
The patent applies preliminary action by using the artifact removal system to pre-process images before diagnostic analysis. By removing both global and local artifacts in advance, the system ensures that diagnosticians receive clean, high-quality images the first time, eliminating the need for re-imaging and thereby reducing patient exposure and stress while maintaining diagnostic accuracy.
3Measurement precision
If artifact removal processing is applied, then image quality is improved, but processing time increases
Solution Approach 1:
The patent segments the artifact removal process into parallelizable stages: global artifact removal, local artifact removal, and high-resolution reconstruction. Each stage can be processed independently and potentially in parallel, reducing overall processing time while maintaining comprehensive artifact removal and high image quality.
Solution Approach 2:
By processing images at multiple resolution levels rather than a single high resolution, the patent reduces computational complexity. Low-resolution processing for global artifacts is faster, and only specific local regions require high-resolution processing, thereby reducing total processing time while maintaining overall image quality.
Data Source
AI summary
An embodiment of the invention may include a method, computer program product and computer system for image artifact removal. The method, computer program product and computer system may include computing device which may receive a primary image and analyze the primary image for global artifacts and local artifacts. The computing device may, in response to identifying a global artifact in the primary image, generate a secondary image with the global artifact removed utilizing a first generative adversarial network. The computing device may, in response to identifying a local artifact in the primary image, generate a patch with the local artifact removed utilizing a second generative adversarial network. The computing device may generate a hybrid image containing a reduction of global artifacts and a reduction of local artifacts by combining the secondary image and the patch utilizing a hybrid generative adversarial network.