Iterative Image Reconstruction Using Chi-Square-Gamma Stop-Criterion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Iterative image reconstruction methods, such as the OSEM algorithm, face challenges in determining the optimal number of iterations, leading to incomplete reconstructions or excessive computational time and artifacts, especially in nuclear imaging where low count rates limit data quality.
Innovation Solution
The use of a chi-square-gamma statistic as a stop-criterion to automatically control the number of iterations in iterative reconstruction algorithms, ensuring that the reconstruction process stops when the statistic meets predefined threshold conditions, thereby optimizing image quality and reducing computational burden.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If the number of iterations is set high to ensure complete reconstruction, then image quality improves, but computational time increases and artifacts may appear
Solution Approach 1:
The patent implements a feedback mechanism by calculating the chi-square-gamma statistic after each iteration and using it to dynamically determine whether to continue or stop the reconstruction process. This feedback loop allows the system to automatically adapt the number of iterations based on actual convergence behavior rather than using a fixed predetermined value, thereby resolving the contradiction between achieving complete reconstruction and minimizing computational time.
Solution Approach 2:
The reconstruction algorithm performs self-evaluation through the chi-square-gamma statistic calculation, allowing it to autonomously determine its own stopping point without external intervention. The system serves itself by monitoring its own convergence status and making decisions about when to terminate iterations, thus optimizing both image quality and computational efficiency simultaneously.
2Productivity
If the number of iterations is set low to reduce computational time, then processing speed improves, but reconstruction becomes incomplete and resolution is lost
Solution Approach 1:
The chi-square-gamma statistic provides continuous feedback during the iteration process, enabling the system to monitor reconstruction progress in real-time. This feedback mechanism ensures that the algorithm stops at the optimal point where sufficient reconstruction quality is achieved, preventing both premature termination and excessive computation, thus resolving the contradiction between processing speed and reconstruction completeness.
Solution Approach 2:
The patent changes the parameter used for stopping criteria from a fixed iteration count to a dynamic statistic-based criterion (chi-square-gamma). This parameter change allows the system to adaptively determine the number of iterations needed, ensuring adequate reconstruction quality while avoiding unnecessary computational steps, thereby balancing processing speed and reconstruction completeness.
3Ease of operation
If a fixed number of iterations is used based on experimentation, then the process is simple to implement, but the result depends on data-specific tuning and may not generalize well
Solution Approach 1:
The reconstruction algorithm becomes self-sufficient by incorporating the chi-square-gamma statistic calculation, which automatically adapts to different data sets without requiring external tuning or experimentation. The system determines its own optimal stopping point based on the specific characteristics of each data set, making it universally applicable while maintaining simplicity in implementation.
Solution Approach 2:
By changing from a fixed iteration parameter (determined through experimentation) to a dynamic statistic-based parameter (chi-square-gamma), the system achieves both simplicity and adaptability. The new parameter automatically adjusts to different data characteristics without requiring manual tuning, thus resolving the contradiction between ease of implementation and data independence.
Data Source
AI summary
An iterative reconstruction method to reconstruct an object includes determining, in a series of iteration steps, updated objects, wherein each iteration step includes determining a data model from an input object, and determining a stop-criterion of the data model on the basis of a chi-square-gamma statistic. The method further includes determining that the stop-criterion of the data model has transitioned from being outside the limitation of a preset threshold value to being inside the limitation, ending the iterations, and selecting one of the updated objects to be the reconstructed object.