Image Cumulative Distribution Function for Tomographic Reconstruction Quality Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current tomographic reconstruction technologies face challenges in evaluating image quality due to the piecewise constant nature of the Poisson cumulative distribution function (CDF), which leads to sensitivity issues and lack of meaningful insights into disparities between predicted and actual results in image space, especially when processed one projection at a time and in data space.
Innovation Solution
Transforming the CDF into an image cumulative distribution function (ICDF) in object space, representing the number of standard deviations associated with each voxel of the image object, allowing for quality control and anomaly detection directly in image space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the Poisson CDF is used for quality control, then the measurement is well behaved at low counts, but the piecewise constant nature causes sensitivity problems and lacks continuity
Solution Approach 1:
The patent transforms the Poisson CDF from its original discrete form to a continuous modified version by adding a random component. This parameter change converts the piecewise constant function into a continuous function that maintains reliability at low counts while eliminating sensitivity problems and providing meaningful gradient information for quality control.
2Stability of the object's composition
If the MCDF is processed one projection at a time in data space, then continuity is achieved, but sensitivity problems occur and meaningful insights into image space disparities are lost
Solution Approach 1:
The patent performs a backprojection operation that transforms the continuous CDF from data space to image space. This dimensional transformation allows the quality control metric to be evaluated in the same space as the reconstructed image, providing meaningful insights into image space disparities while maintaining the continuity benefits of the modified CDF.
3Ease of manufacture
If the CDF is computed in data space, then processing is simplified, but meaningful insights into disparities in image space are not provided
Solution Approach 1:
The patent uses backprojection as an intermediary operation that connects data space and image space. The continuous CDF is first computed in data space (maintaining processing simplicity), then backprojected to image space (providing meaningful insights). This intermediary step bridges the gap between computational ease and diagnostic value.
Data Source
AI summary
Methods and apparatuses for quality control in image space for processing with an input data set are disclosed. A method includes providing an image object, including multiple voxels, and an input data set. A data model is determined from the image object. A cumulative distribution function (CDF) for the input data set is determined from the data model and the input data set based on a plurality of projections. The CDF is transformed to an image cumulative distribution function (ICDF) in object space. The ICDF represents a number of standard deviations associated with each voxel of the image object. The output of the ICDF is displayed. A nuclear imaging system and a computer readable storage medium are also disclosed. Techniques disclosed herein facilitate efficient quality control for tomographic image reconstruction.