Cardiac Cone Beam CT Streak Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cardiac cone beam imaging is plagued by streak artifacts due to excessive x-ray quantum noise, which degrades image quality and is not effectively addressed by existing adaptive filters that do not consider the phase selective nature of the reconstruction algorithm and non-uniform data distribution.
Innovation Solution
The system re-bins projection data into a parallel ray format, filters it using multiple channels with different noise reduction factors, performs weighted averaging of readings contributing to each voxel, and weights the convolved data based on determined noise reduction factors to suppress streak artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If adaptive filters are applied to measured attenuation profiles, then x-ray quantum noise is reduced, but the phase selective nature of cardiac reconstruction and non-uniform data distribution are not considered leading to incomplete artifact suppression
Solution Approach 1:
The patent segments the filtering process into multiple filter channels, each with different noise reduction factors. This segmentation allows the system to handle different noise levels in different angular regions while maintaining adaptability to the phase selective reconstruction algorithm's non-uniform data distribution requirements.
Solution Approach 2:
The patent implements dynamic filtering by determining actual noise reduction factors based on the specific angular distribution of data points contributing to each voxel. This dynamic approach allows the filter to adapt to the varying noise characteristics at different angles and positions, resolving the contradiction between noise reduction and adaptability to phase selective reconstruction.
2Object-affected harmful factors
If multiple filter channels with different noise reduction factors are used, then streak artifacts are suppressed, but computational complexity increases
Solution Approach 1:
The patent performs preliminary actions by determining the actual noise reduction factors based on the weighted averaging of data points before the final reconstruction. This preliminary determination of noise characteristics allows the system to suppress streak artifacts effectively while managing computational complexity through pre-calculated weighting factors.
Solution Approach 2:
The patent changes parameters by using multiple filter channels with different noise reduction factors and dynamically selecting appropriate factors based on angular data distribution. This parameter-based approach suppresses streak artifacts while controlling computational complexity through systematic parameter variation rather than complex algorithmic changes.
3Measurement precision
If phase selective algorithms are used for cardiac reconstruction, then cardiac phase images are obtained, but non-uniform angular data distribution causes pattern noise streaks
Solution Approach 1:
The patent applies dynamic filtering where the noise reduction factor is determined based on the actual angular distribution of data points contributing to each voxel in the phase selective reconstruction. This dynamic adaptation allows the system to maintain cardiac phase imaging accuracy while suppressing the pattern noise streaks generated by non-uniform data distribution.
Solution Approach 2:
The patent implements feedback by using the determined actual noise reduction factors to weight the convolved data from different filter channels. This feedback mechanism ensures that the filtering process adapts to the specific noise characteristics of each angular region, suppressing pattern noise streaks while preserving the accuracy of cardiac phase images.
Data Source
AI summary
In a diagnostic imaging system (10) two-dimensional projection data is collected in a data memory (30). The data is sorted into the data sets collected during selected cardiac phases. A re-binning processor (38) re-bins the projection data into a parallel ray format. An adaptive filter (70) filters the parallel ray format data with each of a plurality of different filter channels based on a calculated photonic noise of each reading and assuming an arbitrary noise reduction factor, different for each filter channel. A convolver (78) convolves the data filtered with each of the filter channels. A noise reduction factor processor (84) determines the actual noise reduction factor occurring due to the weighted averaging of readings contributing to each voxel ν at an angle θε[0, 7π). A weighting processor (90) weights the convolved data from each of the filter channels based on the channels' noise reduction factors and the actual noise reduction factor. A backprojector (98) backprojects the combined weighted sum of projections to build an image representation that can be displayed in a human readable format.