Automatic Transfer Function Determination for CT Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional CT reconstruction techniques face challenges in accurately determining total attenuation coefficients for objects made of multiple materials, leading to undesirable artifacts in 3D representations due to assumptions about theoretical relationships between detector readings and attenuation coefficients, and require precise path length measurements which are difficult to achieve.
Innovation Solution
A computer-implemented method and system that perform one or more test CT reconstructions using different test transfer functions to determine a final photon-count to penetrated depth function, allowing for automatic determination of a real transfer function by selecting the best parameters through non-linear optimization techniques, and using down-sampling techniques to reduce data input and enhance reconstruction speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional theoretical relationship assumptions are used between detector readings and attenuation coefficients, then the CT reconstruction process is simplified, but undesirable artifacts appear in the 3D representation particularly for multi-material objects
Solution Approach 1:
The patent changes the parameter relationship by introducing a learned transfer function that maps detector readings to attenuation coefficients through non-linear optimization, replacing the conventional direct theoretical assumption. This allows the system to adapt parameters based on actual object characteristics rather than fixed theoretical models, resolving the contradiction between simplified processing and accurate representation.
Solution Approach 2:
The system performs self-calibration by automatically determining the transfer function through test reconstructions and optimization without requiring external calibration objects or manual intervention. This self-service approach eliminates the need for complex calibration procedures while maintaining high accuracy for multi-material objects.
2Manufacturing precision
If precise path length measurements are performed for accurate attenuation coefficient determination, then reconstruction accuracy improves, but measurement difficulty increases due to variability in scanned object shape
Solution Approach 1:
The patent replaces the mechanical measurement approach (physical path length measurement) with a computational approach using learned transfer functions and non-linear optimization. Instead of physically measuring path lengths through complex geometric calculations, the system uses iterative optimization to determine attenuation coefficients directly from detector readings, eliminating the measurement difficulty while maintaining accuracy.
Solution Approach 2:
The system transforms the problem from measuring physical path lengths to optimizing parameter relationships through the transfer function. By changing from a geometric measurement parameter to a learned functional relationship, the system avoids the difficulties of precise physical measurement while achieving accurate attenuation coefficient determination.
3Manufacturing precision
If calibration objects of simple shape are developed for every single material, then accurate transfer function determination is achieved, but device complexity and preparation time increase
Solution Approach 1:
The patent creates a universal calibration approach where a single calibration object with diverse materials can be used to determine transfer functions for all materials simultaneously. The non-linear optimization process learns material-specific characteristics from the calibration data, making the calibration system multi-functional rather than requiring separate calibration objects for each material.
Solution Approach 2:
The system uses a calibration object that contains representative samples of various materials, creating a comprehensive reference model. This single calibration object serves as a template from which transfer functions for all scanned materials can be derived through optimization, eliminating the need to physically create separate calibration objects for each material type.
4Manufacturing precision
If multiple test transfer functions are evaluated through test CT reconstructions, then the final transfer function accuracy improves, but processing time increases
Solution Approach 1:
The patent applies partial action by performing test reconstructions with a limited set of candidate transfer functions rather than exhaustively testing all possible functions. The non-linear optimization process efficiently evaluates and compares a manageable number of test cases to identify the best transfer function, achieving high accuracy without excessive processing time.
Solution Approach 2:
The system dynamically selects and evaluates transfer functions based on the specific characteristics of the scanned object. Rather than using a fixed exhaustive search, the optimization process adapts the testing sequence and criteria based on intermediate results, efficiently converging on the optimal transfer function while minimizing unnecessary computations.
Data Source
AI summary
A computer-implemented method of automatically determining photon-count to penetrated depth function for a CT reconstruction includes performing one or more test CT reconstructions on one or more radiographs, each test CT reconstruction using a test transfer function, to provide one more test reconstructions and determining a final transfer function from the one or more test reconstructions.