A method to remove the projection error in triple-energy radiography with contrast medium

The density-map reconstruction of a radiological contrast medium is affected by noise arising from the background lack of homogeneity (the so-called “projection error”) if images of the medium are collected starting from quasi-monochromatic X-ray beams. This noise, especially for a dual-energy reconstruction algorithm, becomes more significant than the statistical fluctuations of the photon-transmitted flux, dramatically reducing the accuracy and the sensitivity of the reconstruction. In this work, we investigate the efficacy of the triple-energy technique, which is based on the simultaneous acquisition of three monochromatic images of the same target injected with contrast medium. A theoretical analysis allows to estimate the sensitivity and the accuracy of the reconstructed density map compared with the dual-energy one (i.e., the density map reconstructed acquiring only two monochromatic images). To validate the theory, a set of experimental measurements was performed: results show that triple energy drastically reduces the projection errors (from 10 to 60 times smaller than the dual-energy one), making it negligible with respect to the statistical noise.