CT Trans-Axial Truncation Compensation With Progressive FOV Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing CT imaging methods face challenges in accurately compensating for trans-axial data truncation due to patient size, positioning, and limited field-of-view, leading to inaccurate reconstruction and undesirable artifacts.
Innovation Solution
A progressive data estimation method is employed, where the image processing system reconstructs an image with a larger reconstruction field-of-view using projection data from a virtual scan and estimated data, refining pixel values based on subject information and curvature, and iteratively adjusts pixel values until the desired reconstruction field-of-view is reached.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If conventional data-driven truncation compensation is used to estimate truncated data beyond the scan FOV, then the reconstruction FOV can be enlarged, but the estimation accuracy decreases with increasing distance from the truncated edge and undesirable artifacts appear in the reconstructed band
Solution Approach 1:
The patent divides the reconstruction FOV into multiple zones: the scan FOV region with high-quality measured data, and the reconstructed band region with estimated data. By segmenting the reconstruction process and applying different processing strategies to different zones, the method maintains high accuracy in the scan FOV while improving accuracy in the previously problematic reconstructed band through iterative refinement using subject information.
Solution Approach 2:
The patent performs preliminary estimation of truncated data using subject information (such as body surface curvature and anatomical priors) before the final reconstruction. This preliminary action provides an initial estimate that is then refined iteratively, preventing the propagation of large errors that would occur with conventional single-step estimation methods.
2Area of stationary object
If hardware enhancements are used to increase the effective imaging FOV (e.g., increasing bore size and detector size), then trans-axial truncation is reduced, but device complexity and cost increase
Solution Approach 1:
The patent replaces mechanical/hardware solutions (larger detectors, larger bore) with a computational/software-based solution. By using subject information and iterative reconstruction algorithms, the method achieves extended FOV reconstruction without requiring physical hardware changes, thereby avoiding increased device complexity and cost.
3Area of stationary object
If multiple scans with different imaging geometry are used to reduce truncation, then the effective FOV is increased, but scanning time and productivity decrease
Solution Approach 1:
The patent performs preliminary estimation of the truncated regions using subject information from the initial scan. This allows the system to reconstruct extended FOV from a single scan without requiring multiple scans with different geometries, thereby maintaining high scanning productivity while achieving the goal of reduced truncation.
Solution Approach 2:
The patent uses subject information (such as body surface curvature and anatomical structures) as a template or copy to estimate and reconstruct the truncated regions. This computational copying approach replaces the need for physical re-scanning with different geometries, significantly reducing scan time while maintaining reconstruction quality.
Data Source
Figure 1A~1B
Figure 2A~2B
Figure 3
AI summary
Imaging apparatuses described herein include a radiation source configured for imaging radiation, a radiation detector positioned to receive radiation from the radiation source, and an image processing system. The image processing system is configured to: receive projection data from the radiation detector, the projection data corresponding to a scan field-of-view (scanFOV), the image being trans-axially truncated; identify a final reconstruction field-of-view (reconFOV) that is larger than the scanFOV; reconstruct an image having a reconstruction field-of-view (reconFOVn), wherein reconFOVn is less than reconFOV; generate a progressive refinement for the image; reproject the image with the progressive refinement thereby generating a virtual scan vscanFOVn; refine the virtual scan vscanFOVn data; and repeat the reconstruction, generation of the progressive refinement, and reprojection for one or more subsequent reconstructions until reconFOV is reached.