Purpose: To study the accuracy of proton stopping power ratio (SPR) determination with dual-energy computed tomography (DECT) for small structures and bone-tissue-air interfaces.
Methods: Hollow cylindrical polylactic acid (PLA) plugs (3 cm diameter, 5 cm height) were 3D-printed containing either one or three septa, thicknesses tsepta = 0.8 mm, 1.6 mm, 3.2 mm and 6.4 mm, running along the length of plug. First, the cylinders were inserted into a tissue equivalent head phantom and DECT scans were obtained using a Siemens SOMATOM Definition Edge CT scanner.  Effective atomic number (Zeff) and electron density (e) images were reconstructed from the DECT to produce SPR-CT images of each plug. Second, independent elemental composition analysis of the PLA plastic was used to determine the Zeff and e for calculating the theoretical SPR (SPR-TH) using the Bethe-Bloch equation. Finally, for each plug, a direct measurement of SPR (SPR-DM) was obtained in a clinical proton beam and compared to SPR-CT and SPR-TH.
Results: SPR-CT for PLA agreed with SPR-TH when tsepta ≥ 3 mm (CT slice thicknesses of 0.5 mm, 1.0 mm, and 3.0 mm). The PLA density was found to decrease when tsepta < 3 mm. As tsepta (and density) decreased, the SPR-CT values also decreased, in good agreement with the behavior of SPR-DM.  
Conclusion: The DECT-based method for calculating SPR in thin heterogeneities showed good agreement with both theoretical and directly measured SPR values. The DECT scans accurately predicted the reduction in SPR associated with decreasing density as the septa thicknesses decreased.

Determination of proton stopping power ratio with dual-energy CT in the presence of small structures and tissue-bone-air interfaces