Optimization of grid-less scattering compensation in X-ray imaging: Simulation study

Abstract

Image procession algorithms for compensation of scattered radiation influence in X-ray imaging were proposed, studied and optimized by numerical simulations. The algorithms include scattering estimation by convolution (superposition) technique, estimation of kernel functions by Monte-Carlo (MC) simulations, determination the optimal number and shape of kernel functions and images segmentation. Determination of the kernel functions' shape and number was performed by Monte-Carlo simulation of realistic Zubal phantom and clustering analysis of functions features. Testing study of algorithms for chest images at 75 keV proves that optimal number of kernel functions is 8 that provides contrast enhancement about 3 times without using of anti-scatter grids. Achieved contrast is about 95% of primary image contrast that exceeds contrast enhancements achieved with anti-scatter grids.