Common-Mask Guided 4DCBCT Reconstruction for Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current four-dimensional cone-beam computed tomography (4DCBCT) systems face challenges in reconstructing high-quality images due to insufficient projection data, leading to severe streaking artifacts and impaired image quality, especially when using the FDK reconstruction algorithm.
Innovation Solution
The common-mask guided image reconstruction (c-MGIR) algorithm separates moving and static portions of the image using a common-mask, allowing for the utilization of full projection information at each optimization iteration, thereby mitigating under-sampling artifacts and improving image quality without increasing the number of projections or scanning dose.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of X-ray projections is increased to provide sufficient data for each respiratory phase, then image quality improves, but scanning dose increases by multiple folds and total scan time increases
Solution Approach 1:
The patent combines projection data from multiple respiratory phases to reconstruct static anatomical structures. By merging information across phases, the system achieves sufficient data for high-quality reconstruction without requiring an increased number of projections per phase, thereby avoiding proportional increases in scanning dose.
Solution Approach 2:
The patent segments the reconstruction process into two distinct components: static structures (reconstructed using combined multi-phase data) and moving structures (reconstructed using phase-specific data). This segmentation allows each component to be optimized independently, achieving high image quality for static structures without requiring excessive projections that would increase dose.
2Measurement precision
If the number of X-ray projections is increased to provide sufficient data for each respiratory phase, then image quality improves, but total scan time increases due to tube firing rate and image panel update rate bottlenecks
Solution Approach 1:
The patent merges projection data across multiple respiratory phases to reconstruct static anatomical structures. This approach achieves sufficient data coverage for high-quality reconstruction without requiring an increased number of projections per phase, thereby avoiding proportional increases in total scan time despite tube firing rate and image panel update rate constraints.
3Productivity
If conventional FDK reconstruction is applied to phase-binned projections, then reconstruction speed is maintained, but severe streaking artifacts occur due to insufficient projections per phase
Solution Approach 1:
The patent segments the reconstruction problem into static and moving components. Static structures are reconstructed using combined multi-phase projection data, which provides sufficient data coverage and eliminates streaking artifacts. Moving structures are reconstructed using phase-specific data. This segmentation enables high image quality without sacrificing reconstruction speed, as the computationally intensive combination of multi-phase data is performed only for static regions.
Solution Approach 2:
The patent introduces a common mask as an intermediary that identifies and isolates static anatomical structures across respiratory phases. This mask acts as a mediator that directs the reconstruction algorithm to combine multi-phase data specifically for static regions, enabling artifact-free reconstruction without requiring increased projections or sacrificing speed in dynamic regions.
4Measurement precision
If post-processing based techniques are used to enhance 4DCBCT images, then image quality may improve, but the methods become impractical when a priori-reconstructed images provide insufficient anatomical information
Solution Approach 1:
The patent performs preliminary segmentation into static and moving components before the main reconstruction process. By pre-identifying static structures using a common mask and combining their projection data in advance, the system creates a solid foundation that eliminates the need for complex post-processing techniques to recover lost anatomical information, thereby reducing overall algorithm complexity.
Data Source
AI summary
A system for constructing images representing a 4DCT sequence of an object from a plurality of projections taken from a plurality of angles with respect to the object and at a plurality of times, first portions of the object being less static than second portions of the object. The system may comprise a processor configured to: iteratively process projections of the plurality of projections for a plurality of groups, each group comprising projections collected while the first portions of the object are in corresponding locations, with each iteration comprising: reconstructing first image portions representing the first portions of the object, the reconstructing of each first image portion of the first image portions being based on projections in a respective group of the plurality of groups; and reconstructing the second image portion, the second image portion representing the second portions of the object, based on projections in multiple groups of the plurality of groups and on the reconstructed first image portions.