Spectrum Binning Approach for Multi-Material Beam Hardening Correction (MMBHC) in CT

In CT, the nonlinear attenuation characteristics of polychromatic X-rays cause beam hardening artifacts in the reconstructed images. When datasets contain multiple materials in the field of scan, the performance of state-of-the-art beam hardening correction approaches is limited by high computational load and reduced correction efficiency. An image based multi-material beam hardening correction (MMBHC) method is introduced by conserving original reconstruction attenuation in a material density map, and obtaining beam harding artifacts by polychromatic forward projecting of the original CT images. By decomposing the attenuation coefficient into photoelectric and Compton scattering components, the spectrum can be parameterized and presented with several energy bins. In this paper, the MMBHC algorithm is accelerated by using a proposed spectrum binning (SB) method to reduce the required prior knowledge from full spectrum to three energy bins. The approach has a significant computational time reduction by use of material number independent forward projection and parameterized energy spectrum with fewer bins. Moreover, representing the spectrum with a few parameters instead of optimizing full spectrum increase the stability of the MMBHC process. The algorithm has been evaluated with a simulated phantom on 3D cone beam geometry. In comparison to correction results with full spectrum information, the results with optimized three bins spectrum shows comparable artifact reduction efficiency while providing significant reduction on computational time.