A New Statistical Image Reconstruction Algorithm for Polyenergetic X-Ray CT

Abstract

This paper presents a new statistical reconstruction algorithm for X-ray CT. The algorithm is based on Poisson statistics and a physical model that accounts for the measurement nonlinearities caused by energy-dependent attenuation. We model each voxel’s attenuation as a mixture of bone and soft tissue by defining density-dependent tissue fractions, maintaining one unknown per voxel avoiding the need of a pre-segmentation. Rather than requiring the entire X-ray spectrum, the method approximates the 2D beam hardening function corresponding to bone and soft tissue with the 1D function corresponding to water and one or two empirical tuning parameters. Results on simulated human data (NCAT phantom) showed a beam hardening reduction similar to conventional post-processing techniques, but with an improved signal to noise ratio.