Voxel Confidence Measurement in CT Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer tomography methods lack a suitable approach to determine the quality or confidence of individual voxels in volumetric models, leading to misinterpretation and detection issues due to the inability to distinguish artifacts from real data.
Innovation Solution
A method is introduced that involves acquiring real projections, reconstructing images, generating artificial projections, and comparing them to calculate confidence measures for each volume unit, using techniques such as forward projection and probabilistic evidence theory to assess voxel confidence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional reconstruction methods are used, then volumetric data can be obtained, but voxel quality cannot be determined leading to misinterpretation of artifacts
Solution Approach 1:
The patent applies preliminary action by generating artificial projections from the reconstructed volume before performing the final comparison. This allows the system to pre-calculate expected projection values and systematically compare them against actual measured projections, thereby determining voxel confidence values before final interpretation. The artificial projections serve as a reference framework that enables subsequent quality assessment of individual voxels.
Solution Approach 2:
The patent implements feedback by using the comparison between artificial and real projections to generate confidence values that feed back into the volumetric data interpretation process. The confidence values provide information about the reliability of each voxel, allowing the system to adjust its interpretation based on this feedback. This creates a closed-loop system where reconstruction quality information is continuously evaluated and used to improve interpretation accuracy.
2Reliability
If no voxel confidence assessment is performed, then the reconstruction process remains simple, but automated analysis algorithms produce incorrect results
Solution Approach 1:
The patent applies segmentation by dividing the volumetric data into individual voxels and assessing the confidence of each voxel separately. Instead of treating the entire volume as a single unit, the system segments the analysis into discrete voxel-level evaluations. This allows automated analysis algorithms to process each voxel with appropriate confidence weighting, improving overall analysis reliability while maintaining manageable computational complexity through systematic breakdown.
Solution Approach 2:
The patent replaces mechanical/physical measurement systems with computational methods for confidence assessment. Instead of using additional physical sensors or measurement devices to verify voxel quality, the system uses mathematical computations comparing artificial and real projections. This substitution of computational analysis for physical measurement reduces device complexity while maintaining or improving reliability of automated analysis.
Data Source
AI summary
A method for determining confidence values for volume units includes acquiring real projections of an object in a tomography system, reconstructing an image of the object from the real projections, generating artificial projections, for each volume unit comparing the real projections with the artificial projections and generating a confidence measure for each of the volume units.