Metal Artifact Reduction in CT Imaging via Frequency Split
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for reducing metal artifacts in CT image datasets often leave residual artifacts, as they primarily focus on correcting high-frequency components without adequately addressing low-frequency noise and beam hardening effects.
Innovation Solution
A frequency split method is employed, where a weighted summation of high-pass-filtered images with and without metal artifact correction is performed, with weightings dependent on proximity to metal, combining low-pass-filtered metal-artifact-corrected images to generate a results image with reduced artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If known metal artifact correction methods are applied, then metal artifacts are reduced, but residual artifacts remain and image quality is not fully improved
Solution Approach 1:
The patent segments the image processing into different frequency components (low-frequency and high-frequency parts) and applies different correction strategies to each. The low-frequency part is corrected using known metal artifact reduction methods, while the high-frequency part is processed separately and then combined, allowing residual artifacts to be eliminated while preserving image details.
Solution Approach 2:
The patent applies local quality by using a weighting function that adapts to local image characteristics. The weighting function assigns different weights to different regions based on their distance from metal artifacts, allowing targeted correction in artifact-prone regions while preserving quality in unaffected regions.
2Measurement precision
If high-frequency components are corrected, then edge details are improved, but low-frequency noise and beam hardening effects are not adequately addressed
Solution Approach 1:
The patent explicitly segments the image into low-frequency and high-frequency components using Fourier transformation. This allows independent processing of each frequency range - low-frequency components are corrected for beam hardening and noise, while high-frequency components preserve edge details, and both are combined to achieve comprehensive image quality improvement.
Data Source
AI summary
A method is disclosed for reducing metal artifacts in CT image datasets. An embodiment of the method includes reconstructing a first CT image dataset with and a second CT image dataset without metal artifact correction, weighted summation of a high-pass-filtered first and a high-pass-filtered second CT image dataset plus a low-pass-filtered second CT image dataset, wherein the weightings are dependent on the proximity to metal in the CT image datasets. A computing unit, a CT system and a C-arm system designed to execute the method are also disclosed.