Iterative Image Reconstruction Using Random Probe Positioning

The patent replaces the mechanical positioning system with a computational approach. Instead of relying on precise physical movement of the radiation source or object to multiple positions, the method uses a single measurement at the object's actual position and computationally reconstructs what the measurement would have been at any other position. This substitution of mechanical movement with computational reconstruction eliminates positioning errors and periodic artefacts while maintaining the ability to generate image data from multiple virtual angles.

The patent creates virtual copies of measurements by computationally calculating what the scattered radiation intensity would be at positions other than where the object is actually located. By using the measured scattered radiation pattern and mathematical models of radiation scattering, the system generates synthetic measurement data for multiple probe positions without physically moving the object or radiation source, thereby avoiding positioning inaccuracies.

The patent performs preliminary computational setup by pre-calculating scattering patterns and establishing mathematical relationships between scattered radiation measurements and object characteristics before the actual iterative reconstruction begins. This preliminary preparation includes modeling the radiation scattering behavior and setting up the computational framework, which accelerates the subsequent iterative process by reducing the computational burden during convergence.

The patent implements an iterative reconstruction process where each iteration uses feedback from previous measurements and calculations to progressively improve the image estimate. The method compares reconstructed image predictions with actual scattered radiation measurements and adjusts the reconstruction parameters accordingly, with the computational efficiency enhanced by the preliminary setup that avoids redundant calculations in each iteration.