Adaptive Projection Matrix for CBCT Recalibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cone-beam computed tomography (CBCT) requires high reproducibility of image chain and movement of imaging apparatus components, leading to time-consuming and costly recalibration when system components are changed, limiting flexibility in capturing projection maps and 3D reconstruction.
Innovation Solution
A computer-implemented method using a trained function to adapt projection matrices based on initial matrices and image quality metrics, allowing flexible capture of projection maps and improved 3D reconstruction by applying a trained neural network to input data from a medical X-ray device, incorporating static and dynamic models of the X-ray device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If offline calibration is performed using a phantom of known geometry to determine projection matrices, then 3D reconstruction can be performed from projection maps, but high reproducibility of image chain and movement of imaging apparatus components is required, leading to time-consuming and costly recalibration when system components are changed
Solution Approach 1:
The system performs preliminary calibration to create a static model of the X-ray device, storing projection matrices in a database for later use. This preliminary action eliminates the need for repeated calibration when system components are not changed, saving time while maintaining precision for stable systems
Solution Approach 2:
The system uses a trained neural network that adapts projection matrices based on feedback from image quality metrics and consistency metrics. This feedback mechanism allows automatic adjustment when system components change, eliminating manual recalibration time while maintaining measurement precision
2Productivity
If offline calibration is performed using a phantom of known geometry, then projection matrices can be stored in a database for later use, but calibration must be repeated from scratch when system components are changed
Solution Approach 1:
The system performs preliminary calibration to create a static model and populate a database with projection matrices, enabling efficient 3D reconstruction for stable systems while maintaining the capability to adapt when changes occur
Solution Approach 2:
The system transitions from static projection matrices to dynamic adaptation using a trained neural network that automatically adjusts projection matrices based on current system state, allowing the system to maintain both efficiency and adaptability simultaneously
3Measurement precision
If 3D capture is bound to a calibrated isocenter position and calibrated capture trajectory, then consistent 3D reconstruction can be achieved, but flexibility in capturing projection maps is limited
Solution Approach 1:
The system replaces fixed calibrated trajectories with dynamic adaptation through a trained neural network that adjusts projection matrices based on actual capture conditions, enabling flexible capture while maintaining reconstruction consistency through automatic correction
Solution Approach 2:
The system changes the approach from fixing geometric parameters (isocenter position, trajectory) to adapting projection matrix parameters dynamically based on image quality and consistency metrics, achieving both flexibility and precision
Data Source
AI summary
A computer-implemented method for providing a three-dimensional (3D) results data set includes: acquiring projection maps of an object under examination which are captured from various projection directions by a medical X-ray device; providing an initial projection matrix based on a static model of the X-ray device; providing a further projection matrix by applying a trained function to input data, wherein the input data is based on the initial projection matrix and the projection maps, wherein at least one parameter of the trained function is adapted based on an image quality metric and/or a consistency metric, and wherein the further projection matrix is provided as output data of the trained function; and providing the 3D results data set through reconstruction from the projection maps by the further projection matrix.