Using Clustering Analysis for Determination of Scattering Kernels in X-ray Imaging

Method for obtaining the optimal number of convolution (superposition) kernels in algorithmic scatter correction for X-ray imaging has been tested using different mathematical phantoms. This method was previously proposed by authors and is based on clustering analysis and realistic Monte-Carlo simulation. Results of the algorithm application were compared to anti-scatter grid application results using realistic simulation of physical processes in X-ray imaging system. Optimal determination of kernel function provides 15-20% contrast increase relative to the uncorrected image that is 10% better than the results of the anti-scatter grid application.