A Hybrid Density-Constrained Dynamic Framework for Five-Material Decomposition and Quantitative Physical Properties Via Multi-Energy Spectral Analysis In Photon-Counting CT
Abstract
Purpose
To develop and validate a novel hybrid framework for five-material decomposition (water, fat, bone, iodine, and lung) using three-energy Photon-Counting CT (PCCT). By integrating multi-energy spectral analysis with density-constrained dynamic basis selection, this method addresses the mathematical underdetermination involved in resolving five materials from three energy measurements, enabling accurate estimation of mass density (MD), effective atomic number (EAN), and stopping power ratio (SPR).
Methods
Spectral images were acquired on a PCCT system (NAEOTOM Alpha, Siemens Healthineers) at 40, 70, and 190 keV. The proposed algorithm combines spectral attenuation analysis with physical density constraints through three adaptive stages: (1) Lung separation—low-density voxels ( 1.30 g/cm³ are constrained to exclude iodine to avoid cortical bone misclassification. Validation was performed using both an electron density phantom (062M, CIRS) and a multi-energy phantom (Sun Nuclear) with iodine inserts (0.2–20 mg/ml).
Results
The framework achieved accurate five-material decomposition. In the 062M phantom, soft tissues (Adipose, Muscle, Liver, Breast 50/50) showed MD errors raging from −0.69% to −2.11% and SPR errors from −0.55% to −2.74% relative to theoretical values. The density constraint effectively suppressed false iodine in high-density bone (Bone 800; MD: +0.24%, SPR: +0.46%). Lung (Inhale) exhibited an MD underestimation of −5.38%. Iodine quantification demonstrated excellent linearity (R² ≈ 1.0) down to 0.2 mg/ml.
Conclusion
The hybrid density-constrained dynamic framework enables robust five-material decomposition using PCCT spectral data, and yielding accurate physical parameters for advanced applications in proton/heavy-ion therapy as well as quantitative imaging.