Material-dependent effects of x-ray pre-filtration on material decomposition in photon-counting CT

Abstract

Objective. Characterize the influence of x-ray source filtration on material decomposition (MD) performance in photon-counting computed tomography (PCCT), understanding effects across different filter materials and thicknesses using contrast agents spanning low-Z (iodine, barium) and high-Z (tantalum) compositions. Approach. Single-contrast phantoms containing iodine, barium, or tantalum, and a dual-contrast phantom containing iodine and tantalum, were scanned on a CdZnTe PCCT system using several pre-filtration conditions: no pre-filter, 0.375 mm brass (single-contrast), and no pre-filter, 0.3 mm Cu, 2.1 mm Cu, 0.15 mm Sn, and 0.6 mm Sn (dual-contrast). MD basis map performance was quantified using voxel-level sensitivity, specificity, quantification accuracy, and contrast-to-pooled-noise ratio. Visual assessments (2D/3D axial slices, histograms) provided complementary analysis of material identification and quantification (MIQ). Main Results. Pre-filtration produced material dependent effects. For iodine and barium, brass filtration improved sensitivity at low concentrations, but increased crosstalk with PMMA and water, reducing specificity. For tantalum, brass filtration improved sensitivity (up to >60-fold increase) and quantification accuracy with increased crosstalk. In dual-contrast imaging, thin and thick Cu and thick Sn improved tantalum sensitivity and quantification, whereas iodine sensitivity and quantification remained stable. Thick Cu and Sn severely reduced specificity for both elements. Across experiments, linear regressions of measured versus ground truth concentrations remained linear (R2 >0.965) and confirmed detector response. Significance. These results demonstrate that optimal filtration strategies for PCCT are material dependent. Moderate filtration enhances MIQ for high-Z agents such as tantalum while preserving specificity, whereas filtration can degrade specificity for low-Z agents. The findings provide guidance for filter selection at PCCT MD applications, particularly those involving tantalum contrast agents, dual-contrast protocols pairing low and high-Z materials, or high-energy virtual monoenergetic image reconstructions for improved visualization of large body habitus. This work establishes a foundation for filtration PCCT protocol design and supports clinical translation of high-Z contrast agents.