Prior-Image Reconstruction Decomposition for Feature Attribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Advanced tomographic reconstruction methods often disregard valuable patient-specific prior information, leading to suboptimal image quality and increased radiation dose, especially in scenarios where prior imaging data can significantly contribute to the understanding of patient anatomy.
Innovation Solution
A novel framework that decomposes the estimation of image features into components supported by current and prior data, allowing for a spatial map to quantify contributions and trace features to their source, thereby enhancing the representation of image features and adjusting the strength of prior information in the reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If prior image-based reconstruction methods are used, then data fidelity requirements are reduced and undersampling is accommodated, but accurate representation of image features becomes compromised
Solution Approach 1:
The patent segments the image reconstruction into distinct components: a first component derived from current projection data and a second component derived from prior images. This segmentation allows independent optimization of each component's contribution, enabling accurate feature representation while reducing data fidelity requirements through controlled integration of prior information.
Solution Approach 2:
The patent applies local quality by differentiating the treatment of different image regions through the decomposition framework. By analyzing the contribution of prior images versus current data at different spatial locations, the method can selectively weight prior information where appropriate while maintaining data-driven accuracy where current measurements are reliable, thus resolving the contradiction between reduced data requirements and feature accuracy.
2Productivity
If prior information is incorporated into reconstruction, then imaging speed is improved and radiation dose is reduced, but the source of image features becomes ambiguous
Solution Approach 1:
The patent implements feedback through the decomposition framework that quantifies and tracks the contribution of prior images versus current data to the final reconstruction. This feedback mechanism provides transparency about feature origins, allowing users to understand which features arise from prior information and which from current measurements, thus resolving the ambiguity while maintaining the speed and dose benefits.
Solution Approach 2:
The patent introduces an intermediary decomposition framework that mediates between prior image information and current projection data. This intermediary layer systematically combines both sources while attributing features to their respective origins, enabling fast reconstruction with reduced radiation dose while maintaining clear knowledge of feature sources through the quantitative contribution analysis.
3Device complexity
If general notions of local smoothness are used, then reconstruction is simplified, but patient-specific prior information is disregarded
Solution Approach 1:
The patent merges the simplicity of general smoothness constraints with the specificity of patient prior information through the decomposition framework. The first component captures general smoothness characteristics while the second component incorporates patient-specific prior information, combining both approaches to achieve simple yet personalized reconstruction that respects individual anatomical variations.
Solution Approach 2:
The patent applies preliminary action by incorporating prior image information before the final reconstruction is completed. By pre-processing and integrating patient-specific prior knowledge into the reconstruction framework, the method simplifies the overall reconstruction process while preserving individual anatomical characteristics, avoiding the need for complex iterative optimization from scratch.
Data Source
AI summary
A framework, comprising techniques, process(es), device(s), system(s), combinations thereof, or the like, to analyze propagation of information in prior-image-based reconstruction by decomposing the estimation into distinct components supported by a current data acquisition and by a prior image. Such decomposition can quantify contributions from prior data and current data as a spatial map and/or can trace specific features in an image to a source of at least some of such features.