CT Image Reconstruction Using Orientation-Based Regularization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computed tomography (CT) image reconstruction often results in artifacts such as staircase artifacts, which degrade image quality and affect diagnostic accuracy.
Innovation Solution
A method and system for image reconstruction that involves obtaining image data, determining local orientation information, and using a regularization item based on gradient and structure tensor information to smooth and modify the image data, reducing artifacts through iterative statistical reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional CT image reconstruction is used, then the reconstruction process is simple and fast, but artifacts such as staircase artifacts appear and image quality deteriorates
Solution Approach 1:
The patent changes the parameters of the regularization term by introducing orientation information through structure tensors. The regularization parameter is dynamically adjusted based on local image orientation characteristics, transforming the reconstruction process to adaptively control smoothing strength in different directions, thereby reducing staircase artifacts while maintaining edge sharpness.
Solution Approach 2:
The patent introduces dynamic adaptation in the reconstruction process by calculating orientation information from structure tensors at each iteration. The regularization term dynamically adjusts its behavior based on local image features, making the reconstruction process adaptive rather than static, which improves image quality without requiring fundamentally complex additional hardware.
2Manufacturing precision
If regularization is applied to reduce artifacts, then image quality improves, but the reconstruction process becomes more complex
Solution Approach 1:
The patent applies local quality by making the regularization term orientation-dependent through structure tensors. Different regions of the image receive different regularization treatment based on their local orientation characteristics. Edges and boundaries are preserved with stronger orientation alignment while smooth regions receive appropriate smoothing, achieving localized optimization of image quality.
Solution Approach 2:
The patent performs preliminary calculation of structure tensors and orientation information before applying the regularization term in the iterative reconstruction. This preliminary action prepares the orientation-dependent parameters in advance, allowing the main reconstruction loop to efficiently use pre-computed orientation information, thereby reducing the overall computational burden despite the enhanced regularization complexity.
3Manufacturing precision
If iterative statistical reconstruction with orientation-based regularization is used, then artifacts are reduced and image quality improves, but the reconstruction time increases
Solution Approach 1:
The patent maintains continuity of useful action by integrating orientation-based regularization into the iterative reconstruction framework. Rather than performing separate processing steps, the orientation information is continuously updated and utilized within each iteration of the statistical reconstruction, ensuring that the beneficial regularization effect is maintained throughout the reconstruction process without requiring additional discrete time-consuming steps.
Solution Approach 2:
The patent applies partial action by selectively applying orientation-dependent regularization only where needed based on local image characteristics. The structure tensor analysis identifies regions requiring orientation-aware treatment (edges, boundaries) versus regions where standard regularization suffices (smooth areas), applying the computationally intensive orientation-based approach only partially to the most critical regions, thereby balancing image quality improvement with computational efficiency.
Data Source
AI summary
The present disclosure relates to systems and methods for image reconstruction. The systems may perform the methods to obtain image data, at least a portion of the image data relating to a region of interest (ROI); determine local information of the image data, the local information relating to orientation information of the image data; determine a regularization item based on the local information; and modify the image data based on the regularization item.