Dose–Time–Quality Relationships In Spectral Photon‑Counting CT across Five Energy Bins
Abstract
Purpose
To determine the optimal balance between radiation dose, acquisition time, and diagnostic performance in spectral photon-counting CT (SPCCT).
Methods
DLP was measured using a CTDI phantom and a (RaySafeTM) Solo ionization chamber. Dose measurements were acquired at 115 kVp and 40 µA across four projection settings (Np = 373, 727, 981, and 1440). For each Np configuration, the corresponding scan time was recorded to evaluate its impact on acquisition duration. Image quality for each Np setting was characterized using data acquired from a homogeneous region of a QRM phantom, reconstructed into five spectral energy bins: E1 = 7–40 keV, E2 = 40–50 keV, E3 = 50–60 keV, E4 = 60–79 keV, and E5 = 79–115 keV. The evaluation metrics—peak signal‑to‑noise ratio (PSNR), root‑mean‑square error (RMSE), and structural similarity index (SSIM)—were computed by comparing the reconstructed images at each dose level with a high‑dose reference dataset. All experiments were performed using an SPCCT system (MARS Microlab 5×120).
Results
Increasing the number of projections resulted in increasing both the DLP and acquisition time up to 141.15 mGy.cm (increase by 322%) and 448.1 seconds (by 272%) from projection 373 to 1440. The improvement of image quality from E1 to E5 is consistent but modest, with a slight increase at higher dose values. We also find that increasing the number of projections resulted in an approximate 3.4% increase in PSNR, along with a 5.2% reduction in RMSE and a 4% improvement in SSIM across the energy bins.
Conclusion
Increasing Np raises dose and scan time with only minor image‑quality gains across the five energy bins, underscoring the need to optimize acquisition for a balanced trade‑off between diagnostic performance, patient safety, and workflow efficiency.