Spatially Varying Weighting Functions for Cone Beam Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cardiac CT scans using low pitch acquisitions and cone beam geometry suffer from artifacts due to incomplete data coverage, leading to distorted structures and shading in reconstructed images, particularly near the x-ray source where data coverage is narrower.
Innovation Solution
A method is introduced to combine images from multiple heart cycles using spatially varying weighting functions based on the x-ray tube position, allowing for the formation of a weighted image that reduces cone beam artifacts by prioritizing data from areas closer to the detector and minimizing extrapolation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If low pitch acquisitions are used to avoid extreme edge of detectors, then detector usage is optimized and radiation dose is reduced, but cone beam artifacts occur due to incomplete data coverage
Solution Approach 1:
The patent combines image data from multiple heart cycles to reconstruct a complete image. By merging data from different cardiac phases, the system achieves sufficient data coverage for high-quality reconstruction without requiring extreme detector edges or high pitch acquisitions, thus resolving the contradiction between image quality and data coverage.
Solution Approach 2:
The patent performs preliminary sorting and selection of projection data from multiple heart cycles before reconstruction. By pre-organizing the data and identifying the most suitable projections in advance, the system ensures adequate data coverage is achieved through intelligent data selection rather than through hardware extremes or high radiation doses.
2Volume of stationary object
If data is extrapolated near the source to achieve complete coverage, then volume coverage is improved, but artifact level increases significantly
Solution Approach 1:
The patent uses the system's own redundant data from multiple heart cycles to fill coverage gaps without external extrapolation. By leveraging the naturally acquired redundant projections from different cardiac phases, the system achieves complete volume coverage using only measured data, eliminating the need for extrapolation and the associated artifacts.
3Productivity
If single sector reconstruction is used for speed, then temporal resolution is maintained, but cone beam artifacts appear in slices closer to detector edge
Solution Approach 1:
The patent segments the reconstruction process into data sorting, weighting function calculation, and weighted combination stages. This segmentation allows efficient processing by handling different aspects separately, maintaining temporal resolution while improving image accuracy through selective data combination and artifact reduction algorithms.
4Quantity of substance
If multiple heart cycles are acquired to improve coverage, then redundant data is available, but combination of data from different cycles introduces complexity
Solution Approach 1:
The patent applies spatially varying weighting functions that assign different weights to data from different heart cycles based on local characteristics. This local quality approach allows the system to utilize redundant data from multiple cycles effectively by adapting the combination strategy to local data quality, thereby managing complexity while maximizing data utilization.
Data Source
AI summary
A method for combining images acquired using helical half-scan imaging comprises identifying an image plane within an overlap region comprising data from first and second view streams representative of first and second cycles of acquired image data. The image plane comprises the same anatomical structure. First and second weighting functions are calculated for first and second images based on first and second tube positions of an x-ray tube. The first and second images correspond to the image plane and are from the first and second view streams, respectively. The first and second tube positions also correspond to the image plane. A weighted image is then formed based on the first and second weighting functions and the first and second images.