Speaker
Description
X-ray spectrum estimation is essential for dose calculation, image quality optimization, and material decomposition in diagnostic radiology. This study implements and evaluates an X-ray spectrum estimation method based on the Birch-Marshall model with three advanced scatter correction techniques. Using a dual-material step wedge phantom composed of acrylic (0-60 mm) and aluminum (0-6 mm), we generated realistic projection images with a tungsten-target X-ray tube across clinical diagnostic energy ranges (40-140 kVp) and a CsI detector model, incorporating Monte Carlo-simulated scatter and quantum noise. We then estimated the spectrum using the Birch-Marshall model combined with different scatter correction methods: Scatter Kernel Superposition Method (SKSM), Statistical Photon Transport Simulation (SPTS), and a proposed Combined Deterministic-Stochastic Approach (CDSA). Performance was evaluated using mean squared error (MSE), normalized root mean squared error (NRMSE), half-value layer (HVL), and mean energy metrics. The CDSA method demonstrated superior performance with 96.34% MSE improvement compared to uncorrected data, followed by SPTS (93.87%) and SKSM (68.25%). Mean energy estimations with CDSA and SPTS (both 40.8 keV) closely matched the reference spectrum (41.2 keV), while SKSM and uncorrected estimates produced identical values of 38.6 keV. Characteristic radiation peaks at tungsten K-shell energies (59.32 keV) were accurately reproduced by CDSA and SPTS methods. These results demonstrate that appropriate scatter correction is crucial for accurate X-ray spectrum estimation, with the proposed CDSA method providing optimal balance between computational efficiency and estimation accuracy.
| Workshop topics | Applications |
|---|
