Iodine quantification performance with deep silicon-based Photon-Counting CT: A virtual imaging trial study

Abstract

Purpose

This study investigates the imaging performance of a deep silicon-based photon-counting CT (Si-PCCT) in quantifying iodine contrast through a virtual imaging trial (VIT).

Methods

We developed a VIT framework using Si-PCCT simulator and benchmarked it against a prototype using an ACR phantom for assessing spatial resolution and noise characteristics, and a geometric phantom for iodine quantification. We imaged geometrical phantoms (20 – 40 cm) with iodine concentrations ranging from 1 to 19.7 mg/ml and XCAT human models with iodine contrast at BMI of 19 to 38 kg/m² across different radiation dose levels (13.9, 27.8, and 41.7 mGy of CTDI_vol). We performed material decomposition, reconstructed iodine CT images, and evaluated iodine quantification accuracy.

Results

The Si-PCCT simulator closely matched with the prototype, with differences within 3 % in MTF (f₅₀ and f₁₀) and 3.7 % (f_peak) in NNPS, and Root-Mean-Square Error of 0.12 mg/ml in iodine quantification. The mean absolute errors (MAE) between the estimated and ground-truth iodine concentration were 0.10, 0.25, and 1.80 mg/ml for 20, 30, and 40 cm phantoms, and 0.31, 0.37, and 0.70 mg/ml for XCAT human models with BMIs of 19, 28, and 38 kg/m², respectively. Similarly, the MAEs were 0.88, 0.45, and 0.31 mg/ml for the geometrical phantoms, and 0.66, 0.5, and 0.46 mg/ml for human models at CTDI_vol of 13.9, 27.8, and 41.7 mGy respectively. These results demonstrate accurate iodine quantification performance, influenced by object size and radiation dose.

Conclusion

This study shows the promising clinical utility of Si-PCCT for accurate iodine quantification under clinically relevant imaging conditions.