X-ray Multi-spectral CT Imaging Method Based On Subtractive Fusion

Multi-spectral CT imaging uses the relationship between the energy spectrum characteristics of the rays and the attenuation coefficient of the material to solve the inconsistency of the reconstruction algorithm of the single energy hypothesis with projection multi-spectrality,which can effectively suppress the hardening artifacts in the CT reconstructed image and improve the image contrast.Existing multi-spectral CT imaging is mainly implemented by multi-energy reconstruction algorithms or photon counting detectors.Multi-energy reconstruction algorithms general y require the material attenuation characteristics or X-ray energy spectrum as a priori,the algorithm model is complex,so the existing methods are difficult to apply to engineering practice.Although the photon counting detector is satisfied the single-energy imaging,but there is a limitation of the detector.Therefore,based on the correlation between the projection sequences under different energy spectra,this paper proposes an X-ray multispectral CT imaging method based on subtraction fusion.Based on the study of fixed voltage CT imaging,the paper first analyze the relationship between multi-energy projections under different energy spectrum distributions and find that there is a strong correlation between the multi-energy projections of the common energy segment corresponding to two different energy spectra,based on this,the relationship between multi-energy projections under the common energy segment corresponding to multiple energy spectrum is studied,and a subtraction fusion model of the variable energy multi-energy projection sequence is established,aiming to remove multiple multi-energy projection common energy segments projection information to obtain projection information with an approximate narrow energy spectrum.In order to solve the problem of small number of photons after subtraction fusion and large reconstruction noise of conventional algorithms,the reconstruction results show that the narrow-spectrum reconstruction problem obtained by subtraction fusion is similar to the low-dose CT reconstruction problem.According to the compressed sensing theory,the image gradient domain as a priori,the statistical iterative algorithm based on total variation minimization(EM-TV)for reconstruction is studied,which can effectively reduce the noise of the reconstructed image.Considering the problem of many iterations of the EM-TV algorithm and residual noise,the EM-TV algorithm based on dictionary learning post-processing is studied.Simulation and practical experiments verify that the algorithm can reduce the number of iterations,improve the convergence speed,and effectively remove the residual noise of the reconstructed image.