X-ray Transmittance Modeling-based Material Decomposition using a Photon-counting Detector CT System

Recently developed x-ray transmittance modeling-based three-step algorithm compensates for the spectral distortion in the photon-counting detector (PCD) and estimates the line-integrals of the basis materials. The x-ray transmittance modeling linearizes the nonlinear forward imaging model and derives a computationally efficient three-step algorithm that achieves almost unbiased and minimum variance estimator. In this article, we apply the algorithm to the experimental data from a research whole-body PCD-computed tomography (CT) system. We perform pixel-by-pixel calibration using water-equivalent phantoms to fit the output of the scanner to the forward model and then feed the calibrated data into the three-step algorithm. Experimental data from iodine phantom and swine abdomen scans demonstrate that the proposed algorithm compensates for the spectral distortion effectively and estimates the line-integrals of two basis materials, water and iodine, efficiently. The proposed method substantially reduces the beam hardening and ring artifacts present in the test phantom compared to those of the image-based material decomposition method and the images directly reconstructed from the raw data of the system. The algorithm also exhibits less bias and comparable noise to those of the other methods for various x-ray energies.