Room: Exhibit Hall | Forum 5
Purpose: This study compares different acquisition modes of dual-energy CT (DECT) for proton therapy planning and their potential to estimate relative stopping powers (RSP) for proton therapy. Acquisition modes different from dual-source DECT (DSDECT) are investigated due to their applicability to radiation therapy. The results of this study are intended to facilitate clinical decision making, especially for proton therapy centres aiming to use DECT in the future.
Methods: Four CT types capable of DECT are investigated: 1) Siemens Somatom Edge (Twin Beam DECT), 2) GE Revolution (rapid kV switching DECT), 3) Toshiba Acquilion One (sequential DECT) and 4) Philips IQon (dual-layer DECT). For each scanner, a Gammex phantom (RMI 467) for calibration purposes and a CIRS phantom for evaluation are scanned. DECT images are acquired with 80 and 140 kV (135 kV for Toshiba) tube voltages. RSP values are estimated voxelwise following the method by Bourque et al. (2014). The estimated phantom RSP values are compared to theoretical values for self-consistency (Gammex phantom) and accuracy (CIRS phantom).
Results: Self consistency of each acquisition mode is reported in the following as root-mean-square error (RMSE) over all investigated materials: 4.30% (Twin Beam), 2.80% (rapid kV switching), 3.25% (Sequential), 0.98% (dual-layer). Accuracy values are: 34.67% (Twin Beam DECT), 1.59% (kV switching), 1.34% (sequential), 3.01% (dual-layer DECT).
Conclusion: In the self-consistency study, only dual-layer DECT produces errors similar to DSDECT (0.67% according to Bourque et al., 2014). In the accuracy study, both rapid kV switching and sequential DECT perform similarly well, followed by dual-layer DECT, while Twin Beam DECT produces large RSP errors. Our results suggest that for proton therapy planning purposes, the use of either a dual-layer DECT or a rapid kV switching is recommended.