Relative Image Quality Ranking for Medical Artifact Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods for artifact reduction in medical image data, such as motion and metal artifacts in X-ray imaging, often converge to local quality optima due to the use of relative image quality metrics, leading to inefficiencies and suboptimal results.
Innovation Solution
A computer-implemented method that orders intermediate data sets by relative image quality using a sorting algorithm, adjusts quality limit values based on input data and acquisition information, and iteratively refines corrections to achieve a global optimum, reducing the likelihood of local optima and ensuring robust convergence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a trained function with high complexity is used to determine absolute image quality measure, then measurement precision is improved, but device complexity and training time increase
Solution Approach 1:
The patent replaces the complex, time-consuming absolute quality metric training with a simpler relative quality metric approach. Instead of training a comprehensive model to determine absolute image quality, the system uses multiple correction approaches and compares them relatively, achieving good results without the heavy training overhead.
Solution Approach 2:
The patent applies multiple correction approaches (more than the single optimal correction) and uses a sorting algorithm to evaluate them. This partial approach of evaluating multiple candidates with relative metrics achieves sufficient precision without requiring the full complexity of absolute quality determination.
2Device complexity
If relative image quality metric is used for optimization, then device complexity is reduced, but manufacturing precision deteriorates due to local optima
Solution Approach 1:
The patent segments the correction optimization into multiple independent correction approaches, each generating an intermediate data set. By dividing the problem into multiple segments (different correction methods) and evaluating them separately through sorting, the system avoids getting trapped in local optima while maintaining computational efficiency.
Solution Approach 2:
The patent introduces dynamic elements by using a sorting algorithm that can adaptively evaluate and reorder correction approaches. The system dynamically selects the best correction based on relative quality metrics, allowing it to escape local optima by exploring multiple correction paths rather than following a single fixed optimization trajectory.
3Measurement precision
If comprehensive training with extensive training data is performed, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent performs preliminary actions by generating multiple intermediate data sets with different correction approaches before the final quality assessment. This preliminary diversification of correction methods allows the system to avoid time-consuming retraining, as the relative comparison approach can be applied immediately to the generated data sets.
Solution Approach 2:
The patent creates multiple copies of the input data set, each processed with a different correction approach. These copied and processed data sets are then evaluated using the sorting algorithm with relative quality metrics, achieving precise quality assessment without the need for extensive training data and time.
4Device complexity
If a single correction approach is used, then device complexity is reduced, but reliability deteriorates due to local quality optima
Solution Approach 1:
The patent applies local quality principles by evaluating each correction approach individually with its own intermediate data set. The sorting algorithm then identifies which local correction approach yields the best relative quality, ensuring reliable optimization by selecting the best local solution rather than relying on a single correction method.
Solution Approach 2:
The patent creates a composite correction system by combining multiple different correction approaches. Instead of using a single correction method, the system integrates multiple correction techniques and uses the sorting algorithm to determine the optimal combination, thereby improving reliability while managing complexity.
Data Source
AI summary
Systems and methods for image processing that provides an output data set on the basis of an input data set relating to an examination subject. The method includes generating an initial intermediate data set group from a plurality of intermediate data sets by applying a respective correction operation associated with the respective intermediate data set to the input data set, ordering the intermediate data sets according to a quality ranking in respect of their image quality by a sorting algorithm, wherein the intermediate data sets of the initial intermediate data set group are ordered in a first iteration, wherein, in the sorting algorithm, a respective relative quality metric for the relative image quality of a respective first of the intermediate data sets in relation to a respective second of the intermediate data sets is determined by a comparison algorithm, after which the order of the respective first and second intermediate data set in the quality ranking is specified by a comparison of the relative quality metric with a quality limit value.