Image Quality Index for CT Low Contrast Detectability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In CT imaging, variability in training and individual preferences among radiologists lead to inconsistent parameter selection, affecting diagnostic quality, especially with de-noising reconstruction algorithms where traditional image quality metrics are misleading due to removal of noise cues, and dose modulation algorithms struggle to maintain consistent image quality across different acquisition conditions.
Innovation Solution
A method that determines low contrast detectability to compute an image quality index, which identifies appropriate acquisition or reconstruction parameters and visually presents this index, ensuring confidence in low contrast object detection, even with de-noising algorithms, by using a clinical indication and pre-determined mappings between image quality indexes and parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If de-noising reconstruction algorithms are used to remove noise from images, then image quality is improved, but visual noise cues that indicate image quality and confidence are lost
Solution Approach 1:
The patent introduces an intermediary metric (noise metric calculated from projection data) that mediates between the raw projection data and the final de-noised image. This metric provides quantitative information about the noise level and image quality without requiring the visual noise cues to be present in the final image, thus resolving the contradiction between image quality improvement and loss of quality information.
2Device complexity
If traditional image quality metrics based on noise are used, then image quality assessment is simplified, but they provide misleading information when used with de-noising reconstruction algorithms
Solution Approach 1:
The patent replaces the traditional mechanical/visual approach to quality assessment (relying on visual noise cues in the final image) with a computational approach (calculating noise metrics from projection data using statistical models). This substitution allows for accurate and consistent image quality assessment that works correctly with de-noising reconstruction algorithms, resolving the contradiction between assessment simplicity and metric accuracy.
3Stability of the object's composition
If dose modulation algorithms target uniform image noise to achieve constant image quality, then image quality consistency is improved, but they fail to account for low contrast detectability changes
Solution Approach 1:
The patent changes the parameters used for image quality assessment from simple noise metrics to comprehensive metrics that include low contrast detectability calculations. By incorporating task-specific parameters (low contrast object detection capability) into the quality assessment, the system can optimize dose modulation to maintain both image quality consistency and adequate low contrast detectability, resolving the contradiction between these two objectives.
Data Source
AI summary
A method includes determining a low contrast detectability of a scan and generating an image quality index based on the determined low contrast detectability. Another method includes identifying an image quality index of interest, identifying an acquisition and/or reconstruction parameter based on the image quality index and a pre-determined mapping between image quality indexes and acquisition parameter and reconstruction parameters, and displaying the identified acquisition and/or the reconstruction parameter. A system (100) includes a metric determiner (122) that determines a first image quality index for a scan based on at least one of a low contrast detectability of the scan or a project domain noise of the scan, and/or a parameter recommender (126) that recommends at least one of an acquisition or a reconstruction parameter for a scan based on a second image quality index, and a display (114) that visually presents the first or second image quality index.