Scatter Correction in Half-Fan CT Using Hybrid Kernel Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current scatter correction methods in CT imaging systems, particularly those using half-fan geometry with an offset detector, underestimate scatter from patient tables due to their shape, position, and material composition, leading to artifacts like cupping in CT images.
Innovation Solution
The implementation of the Piercing Point Equalization method and the Analytical Hybrid Kernel model to estimate and correct scatter from adjacent objects, such as patient tables, by comparing opposing projections and using a hybrid kernel approach that combines symmetric and asymmetric kernels based on projection angles and table geometry.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional scatter correction methods are used in half-fan geometry CT systems, then the imaging process is simple and fast, but scatter from patient tables is underestimated leading to cupping artifacts and reduced image accuracy
Solution Approach 1:
The scatter correction method is segmented into two distinct components: a fast approximate correction applied to all projections, and a specialized correction for table scatter applied only to projections where the table is visible. This segmentation allows the system to maintain overall simplicity while improving accuracy for specific problematic cases.
Solution Approach 2:
The patent applies different scatter correction strategies to different projections based on local conditions. For projections where the patient table is not visible, conventional methods are used. For projections where the table is visible, the specialized hybrid kernel method with asymmetric kernels is applied. This local adaptation of correction quality resolves the contradiction between simplicity and accuracy.
2Productivity
If scatter from patient tables is not corrected, then the processing is faster and simpler, but cupping artifacts appear in CT images reducing diagnostic quality
Solution Approach 1:
The patent applies scatter correction selectively rather than universally. The specialized table scatter correction is applied only to projections where the table is visible in the field of view, determined by geometric calculations. This partial application of correction maintains processing speed while eliminating artifacts in the affected regions.
Solution Approach 2:
The patent introduces an intermediary geometric analysis step that determines which projections require table scatter correction. By calculating whether the patient table intersects with the X-ray beam path for each projection, the system selectively applies correction only where needed, maintaining efficiency while reducing artifacts.
3Measurement precision
If symmetric kernels are used for scatter estimation, then the calculation is simpler, but scatter from adjacent objects like patient tables is not accurately modeled
Solution Approach 1:
The patent employs asymmetric kernels specifically for modeling scatter from the patient table, recognizing that the table's geometry and position create asymmetric scatter patterns. The asymmetric kernel formulation accounts for the directional nature of scatter from the table's surface and edges, providing accurate modeling where symmetric kernels would fail.
Solution Approach 2:
Different kernel types are applied locally based on the scatter source: symmetric kernels are used for scatter from the patient body, while asymmetric kernels are used for scatter from the patient table. This localized application of appropriate kernel complexity achieves accurate scatter estimation without unnecessarily complicating the overall model.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
These methods effectively reduce scatter-related artifacts, improving the accuracy of CT reconstructions and reducing noise-induced fluctuations, resulting in enhanced image quality and corrected pelvis CT slices with improved Hounsfield Units.
Implementation Method 1
transmitting radiation from a 'point source' 105 through the object 102, which will absorb some of the radiation based on its size, density, and atomic composition
Implementation Method 2
when a quantum of radiation is absorbed by a portion of the object, one or more scattered rays are often generated that deviate from the transmission path of the incident radiation
Data Source
AI summary
Embodiments of the disclosure generally set forth techniques for adjusting an estimate of scattered radiation of a target object. One example method includes generating a plurality of radiographic projections, selecting a first radiographic projection and a second radiographic projection of the target object from the plurality of radiographic projections, forming first estimates of scattered radiation in the first and second radiographic projections, applying the first estimates of scattered radiation to the first and second radiographic projections to generate a modified first radiographic projection and a modified second radiographic projection, comparing the modified first and second radiographic projections to generate correction modules, and applying one of the correction modules to a first subset of the plurality of radiographic projections and another of the correction modules to a second subset of the plurality of radiographic projections.