Filtered Back-Projection Artifact Reduction in Tomosynthesis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Tomosynthesis imaging faces significant challenges in reducing cross-talk artifacts due to the restricted set of 2D projections, which are exacerbated by limited angular sweeps and low projection counts, leading to image distortions and obscured details.
Innovation Solution
A computer-implemented method using filtered back-projection that creates a modified filtered back-projection density distribution by emphasizing outlier contributions, calculated through weighted linear combinations of standard and modified filtered back-projection density distributions, to reduce artifacts without requiring hardware upgrades or modifying scanning processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If the number of 2D projections is increased to blur high-frequency cross-talk artifact structures, then image artifacts are reduced, but scan time and computation time increase
Solution Approach 1:
The patent extracts and processes only the high-frequency components of the projection data that cause cross-talk artifacts. By applying a high-pass filter to isolate these problematic frequency components and processing them separately through a corrected back-projection formula, the method removes artifacts without requiring additional projections or increased scan time.
Solution Approach 2:
The patent changes the processing parameters by applying different filtering operations to different frequency components. Standard filtered back-projection is applied to low-frequency components while a modified formula with high-pass filtering is applied to high-frequency components, allowing artifact reduction without increasing the number of projections.
2Object-affected harmful factors
If the number of 2D projections is increased to blur high-frequency cross-talk artifact structures, then image artifacts are reduced, but computation time increases
Solution Approach 1:
The patent segments the reconstruction process into two distinct parts: standard filtered back-projection for the complete set of projections, and a separate correction step that processes only the high-frequency components. This segmentation allows the computationally intensive artifact reduction to be performed on a subset of the data rather than the entire dataset.
Solution Approach 2:
The patent extracts only the high-frequency components that contain the cross-talk artifacts and processes them separately. By applying a high-pass filter to isolate these components and using a corrected back-projection formula only on this extracted subset, the method reduces computation time compared to processing all projections with artifact-reduction algorithms.
3Object-affected harmful factors
If a larger tomosynthesis scan angle is used to reduce data insufficiency, then cross-talk artifacts are reduced, but device complexity and space requirements increase
Solution Approach 1:
The patent replaces the mechanical solution of increasing scan angle with a computational approach. Instead of requiring the scanner to physically move through a larger angular range, the method uses mathematical processing—specifically high-pass filtering and corrected back-projection—to eliminate cross-talk artifacts from data acquired at limited angles.
Solution Approach 2:
The patent changes the processing parameters applied to the projection data rather than changing the physical scanning parameters. By applying different filtering and reconstruction formulas to the existing data, the method achieves artifact reduction without modifying the scanner's mechanical configuration or scan geometry.
Data Source
Figure 1~2
Figure 3~6
AI summary
A computer-implemented method of deriving 3D image data of a reconstruction volume from a plurality of 2D projections (Pi) by means of filtered back-projection, comprises the steps of: - receiving a plurality of 2D projections (Pi) of an imaged object, each 2D projection (Pi) corresponding to a projection plane; - applying a filter to each of the 2D projections (Pi) to yield filtered 2D projections (P̂ι); and - calculating a filtered back-projection density distribution (f) from the filtered 2D projections (P̂ι) by means of filtered back-projection. According to the invention, the computer-implemented method further comprises the steps of: - calculating at least one modified filtered back-projection density distribution (fC1, fC2) that is indicative of outlier values included in the filtered 2D projections (P̂ι) by means of filtered back-projection; and - calculating a revised filtered back-projection density distribution (fimpr) as a weighted linear combination of the filtered back-projection density distribution (f) and the at least one modified filtered back-projection density distribution (fC1, fC2) .