Purpose: To investigate the impact of intravenous (IV) contrast agent on proton therapy dose calculation using virtual non-contrast (VNC) dual-energy CT (DECT) images.
Methods: CT scans acquired on a Siemens Definition Edge scanner are used to assess the impact of IV contrast on stopping power ratios (SPR) predicted by single-energy CT (SECT) and DECT. SECT images are converted to SPR using the stoichiometric method, while two methods are considered for the DECT images: 1) an electron density and effective atomic number (ED-EAN) formalism and 2) a VNC approach. Moreover, two modes are used with DECT: 80/140 kVp sequential acquisitions and 120 kVp Twin-Beam. Experimental measurements of SPR are obtained for tissue-like plastics and different concentrations of IV contrast in water using a multi-layer ion chamber (MLIC) and compared with predictions made by the different imaging modalities. Real patient contrast enhanced CT images of proton therapy patients are used to quantify the impact of IV contrast on dose distributions using all three methods. Results are validated against the dose calculated on a non-contrast SECT scan, used as reference.
Results: Experimental measurements in tissue-like plastics demonstrate the superiority of the Sequential over TwinBeam DECT acquisitions for both approaches. In phantoms with variable IV contrast concentrations, measurements made with the MLIC agrees with Sequential DECT predictions within 2.15% for the ED-EAN method and 1% for VNC, while SECT introduce errors above 20%. Dose distributions calculated in patient images show that IV contrast introduce errors up to 2.6% on range. Using DECT-based VNC images, the range error is reduced below 0.1%.
Conclusion: This study demonstrates that the error introduced by IV contrast in dose calculation for proton therapy can be substantially reduced using DECT based SPR calculation methods and that a single contrast enhanced scan can be used for both contouring and dose calculation.