摘要
To address the lack of standardized benchmarking in spectral micro-CT, this study establishes a comprehensive system-level performance reference for a photon-counting detector (PCD) platform. We jointly evaluated detector spectral characteristics, including X-ray fluorescence-based energy calibration and count-rate behavior, alongside core imaging metrics using standardized phantoms. Notably, the PCD-CT system demonstrated superior spatial resolution and quantitative accuracy compared to conventional energy-integrating detector (EID) systems. The system achieved a 10% MTF of 28.99 lp/mm, attributable to its direct-conversion architecture. Furthermore, NPS analysis confirmed consistent noise texture, while bone mineral density measurements indicated significantly improved accuracy over EID-CT, highlighting its reliability for precision densitometry. Spectral quantification was validated through a projection-based material decomposition pipeline; using an independent hold-out set, the system achieved sub-millimeter thickness estimation (e.g., 0.084 mm RMSE for Al). This framework enabled virtual monochromatic imaging to effectively suppress metal artifacts while preserving structural fidelity. Importantly, these comparisons were conducted without proprietary ring-artifact reduction to isolate the intrinsic algorithmic performance. This work provides a rigorous measurement-centric benchmark essential for the design, optimization, and cross-platform evaluation of next-generation spectral micro-CT systems.