Speaker
Description
Beam-hardening and material cross-contamination remain significant challenges in quantitative computed tomography of multi-component samples acquired using polychromatic X-ray sources. For medical applications, the limited precision and portability of CT scan results often necessitate multiple diagnostic scans, leading to unnecessary radiation exposure for patients.
In this work, a polychromatic Quantitative Absorption Tomography (QAT) framework is extended to reconstruct component-specific density distribution maps in a two-component sample. To mitigate beam-hardening effects, the framework decomposes projection sinograms acquired under two distinct incident X-ray spectra into component-specific column density sinograms using a spectral forward model. Residual cross-contamination artifacts are addressed using a quantitative correction strategy based on spatial reference masks obtained by segmenting conventional linear attenuation reconstructions. Contaminated regions are identified from these masks and reinterpreted in the sinogram domain using the same polychromatic forward model prior to back-projection. Experimental validation was performed on intertwined Al-Cu wire sample. The proposed correction improves material separation, reduces streak artifacts, and brings reconstructed densities within one standard deviation of independently measured reference values. The new technique promises significant improvements in the precision and portability of CT scan results, which are important factors in reducing radiation exposure for medical patients.