Ordered Subsets Momentum for X-ray CT Image Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Iterative reconstruction techniques in CT imaging are computationally expensive and slow, limiting their use in clinical settings due to the need for multiple iterations and long computations, despite providing greater flexibility and image quality improvements.
Innovation Solution
The implementation of Ordered Subsets (OS) with momentum terms, specifically using Nesterov's algorithms, to accelerate the convergence of iterative image reconstruction, reducing computational costs and improving image reconstruction speed by using a subset of projection data and derived momentum terms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If iterative reconstruction techniques are used to improve image quality and flexibility, then imaging metrics such as radiation dose, noise, and artifacts are improved, but computational cost and reconstruction time increase significantly
Solution Approach 1:
The patent divides the projection data into ordered subsets and processes only a subset of data in each iteration, rather than using all data simultaneously. This segmentation approach reduces the computational burden per iteration while maintaining the benefits of iterative reconstruction for improving image quality metrics.
Solution Approach 2:
The patent introduces momentum terms that dynamically adjust the reconstruction process based on previous iterations. The momentum terms capture convergence behavior and adapt the reconstruction steps, enabling faster convergence rates of O(1/(Mk)^2) compared to conventional methods, thus reducing total reconstruction time.
2Productivity
If ordered subsets algorithms are used to accelerate image reconstruction by using only a subset of measured projection data, then computational speed is dramatically improved, but the algorithms still require multiple iterations and involve long computations
Solution Approach 1:
The patent enhances conventional ordered subsets algorithms by introducing momentum terms that dynamically adapt to the convergence behavior of the iterative process. This dynamic enhancement accelerates convergence to O(1/(Mk)^2), significantly reducing the number of iterations needed and total computation time compared to standard OS algorithms.
Solution Approach 2:
The patent changes the convergence parameters by introducing momentum terms that modify the update rules based on historical iteration data. This parameter change transforms the convergence rate from linear to quadratic-like (O(1/(Mk)^2)), dramatically reducing computation time while maintaining reconstruction accuracy.
3Speed
If conventional ordered subsets algorithms are used, then initial acceleration is provided, but the algorithms still employ a number of iterations to converge and involve long computations
Solution Approach 1:
The patent enhances the dynamic behavior of the reconstruction algorithm by introducing momentum terms that capture and exploit convergence patterns from previous iterations. This dynamic enhancement transforms the convergence rate from linear to approximately quadratic (O(1/(Mk)^2)), significantly improving productivity while maintaining the initial acceleration benefits of ordered subsets.
Data Source
AI summary
Methods, systems, and non-transitory computer readable media for image reconstruction are presented. Measured data corresponding to a subject is received. A preliminary image update in a particular iteration is determined based on one or more image variables computed using at least a subset of the measured data in the particular iteration. Additionally, at least one momentum term is determined based on the one or more image variables computed in the particular iteration and/or one or more further image variables computed in one or more iterations preceding the particular iteration. Further, a subsequent image update is determined using the preliminary image update and the momentum term. The preliminary image update and/or the subsequent image update are iteratively computed for a plurality of iterations until one or more termination criteria are satisfied.