Tomosynthesis Reconstruction Using Geometric 3D Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Limited-angle tomographic imaging results in poor depth resolution and blurring of structures, leading to unrealistic reconstructions due to the lack of depth information and convergence issues in iterative reconstruction algorithms.
Innovation Solution
An image processing system that incorporates geometric prior knowledge by forming a 3D model of specified structures within the input image volume, adapting the image volume to include depth information, and using this adapted volume as an initial image for iterative reconstruction to improve depth resolution and reduce artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If limited-angle tomographic imaging is performed to reduce imaging time and complexity, then imaging speed and device simplicity are improved, but depth resolution deteriorates
Solution Approach 1:
The system performs a preliminary reconstruction to generate an initial 3D image volume, then uses this initial volume to guide the formation of geometric models for structures of interest. This preliminary action provides depth information that is then incorporated into the reconstruction process, resolving the depth resolution issue without requiring full-angle imaging.
Solution Approach 2:
Geometric models serve as intermediaries between the limited projection data and the final reconstructed image. These models encode depth information about structures like lesions and calcifications, acting as a bridge that transfers depth knowledge into the reconstruction process without requiring additional projection views.
2Measurement precision
If iterative reconstruction algorithms are used to improve image quality, then measurement precision is improved, but convergence to realistic solutions deteriorates due to multiple consistent solutions
Solution Approach 1:
The system uses the initial reconstructed volume as feedback to guide the iterative reconstruction process. The geometric models derived from this initial volume provide feedback constraints that steer the iterative algorithm toward realistic solutions, preventing convergence to unrealistic alternatives that are mathematically consistent with the projection data.
Solution Approach 2:
The system changes the parameter space by incorporating geometric model parameters (size, shape, position) into the reconstruction process. This adds constraints that reduce the solution space from multiple mathematically valid solutions to a smaller set of physically realistic solutions, improving reliability of convergence.
3Device complexity
If standard initial images are used for iterative reconstruction, then device complexity is reduced, but depth information and contrast-to-noise ratio deteriorate
Solution Approach 1:
Instead of using simple uniform initial images, the system performs a preliminary reconstruction to create an informed initial volume that contains depth information about the imaged object. This preliminary action, while adding some computational steps, provides crucial depth knowledge that improves subsequent iterative reconstruction without requiring complex hardware modifications.
Data Source
AI summary
An Image processing system (IPS) and related method and imaging arrangement (IAR). The system (IPS) comprises an input interface (IN) for receiving i) a 3D input image volume (V) previously reconstructed from projection images (π) of an imaged object (BR) acquired along different projection directions and ii) a specification of an image structure in the input volume (V). A model former (MF) of the system (IPS) is configured to form, based on said specification, a 3D model (m) for said structure in the input 3D image volume. A volume adaptor (VA) of the system (IPS) is configured to adapt, based on said 3D model (m), the input image volume to so form a 3D output image volume (V′).