A Detector Response Model Based Ml-Fm Spectrum Reconstruction Method Calibrated By Xrf Spectra

Ruizhe Li,Liang Li,Zhiqiang Chen

DOI: https://doi.org/10.1109/NSSMIC.2016.8069485

2016-01-01

Abstract:Accurate estimation of distortion-free spectrum is important but difficult in various applications including spectral CT systems. To reconstruct the incident spectrum, there are two key problems to solve. One is the acquisition of detector response function. Different from many existing methods that either require detailed Mont-Carlo modeling of the detector system, or require monochromatic light from synchrotron sources or radioactive isotope sources, in this study we promote a feasible method based on the parametric modeling of X-ray fluorescence spectra, which is inspired by an existing model based on the XRF spectra. The other key problem is reconstructing the spectrum based on the obtained detector response. Since the system is severely ill-conditioned, it is very difficult to obtain an accurate solution from noisy data. Existing method uses a Spectrum Stripping algorithm, whereas with an imperfect assumption that could lead to inaccurate results. We propose to reconstruct the incident spectrum by ML-EM algorithm based on the Poisson noise model of the system. To verify our method, simulation and experiment are conducted with both the ML-EM method and the existing method for comparison. Results show that the proposed method is effective for the reconstruction of incident spectra, especially for XRF spectra, for which our method markedly increases the accuracy comparing to the existing method.