CT Artefact Estimation via Gradient Lookup Tables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large coverage CT imaging often results in insufficient data for exact reconstruction due to cone beam artefacts, which are caused by the cone-shaped x-ray beam, leading to suboptimal medical image quality.
Innovation Solution
A computer-implemented method generates an artefact estimation image by identifying gradients in the CT image, using a look-up table to predict the effect of cone-beam artefacts, allowing for the reduction of these artefacts in the image by subtracting the artefact estimation from the original CT image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If large coverage CT imaging is performed with a cone-shaped x-ray beam, then the coverage area is increased, but cone beam artefacts are generated leading to insufficient data for exact reconstruction
Solution Approach 1:
The invention segments the cone beam artifact correction problem into two distinct phases: first reconstructing the image using conventional algorithms despite artifacts, then separately estimating and removing the artifacts in a second reconstruction pass. This segmentation allows each phase to be optimized independently, resolving the contradiction between maintaining large coverage and achieving accurate reconstruction.
Solution Approach 2:
The method performs preliminary artifact estimation by identifying gradients in the initially reconstructed image before final artifact removal. This preliminary action allows the system to predict artifact locations and characteristics based on the first reconstruction, enabling more accurate artifact subtraction in the second pass while preserving true anatomical features.
2Manufacturing precision
If conventional artifact reduction methods are used, then artefact reduction is achieved, but complex computations like FFT or forward projection are required
Solution Approach 1:
The invention extracts only the essential gradient information from the reconstructed image to drive artifact estimation, rather than performing complete forward projections or FFT operations. By taking out only the gradient magnitude and direction data, the method achieves artifact reduction with significantly reduced computational complexity while maintaining image quality.
Solution Approach 2:
The method creates a simplified copy of the artifact estimation process that mirrors the structure of complex forward projection but uses gradient-based calculations instead. This copying approach preserves the essential artifact modeling functionality while replacing computationally intensive operations with simpler gradient-based estimations, reducing device complexity.
3Device complexity
If gradient-based artifact estimation is performed, then computational complexity is reduced, but only gradients greater than a predetermined value are identified
Solution Approach 1:
The invention applies local quality by differentiating the treatment of gradients based on their magnitude: strong gradients (greater than predetermined value) are processed with full artifact estimation, while weaker gradients are either processed with reduced estimation or ignored. This local differentiation maintains artifact estimation precision for significant features while reducing computational complexity for minor variations.
Solution Approach 2:
The method changes the parameter threshold for gradient processing by only considering gradients greater than a predetermined value. This parameter change filters out noise and minor variations, focusing computational resources on significant anatomical boundaries and artifact sources, thereby maintaining precision where it matters most while reducing overall computational complexity.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
A mechanism for generating an artefact estimation image that represents the effect of cone-beam artefacts in a computed tomography (CT) image. This is achieved by identifying the position of gradients (being sudden changes of intensity) in an axis of the CT image parallel to a rotation axis of the CT system that generated the CT image, where each gradient represents a source of a cone-beam artefact. A look-up table is used to individually identify the effect of a cone-beam artefact on areas surrounding each identified position of the gradient, to generate an artefact estimation image.