Develop and Evaluate a Dose Calculation Strategy Using Electron Density Maps from Spectral CT

Abstract

The conventional method of estimating relative electron density using Hounsfield Units (HUs) is prone to errors resulting from various factors such as energy spectrum, exposure, scanner/patient conditions, etc. Specific calibration is needed for each acquisition protocol. To overcome these limitations, dual energy CT has been extensively researched for its accuracy in dose calculation using tube potential switching techniques. A dual layer design offers a different approach to acquire spectral images using single data acquisitions. This study aims to develop and evaluate a dose calculation method using electron density maps generated directly from a dual layer detector scanner. A phantom with tissue equivalent inserts was scanned using different scanner configurations on a dual layer detector scanner. The electron density of 17 inserts ranged from 0.668 - 5.663 × 1023 m-23. The energy dependent attenuation curves were generated and translated into values of Compton and photoelectric components, which were used to calculate ED values. The ED values were compared to normal values provided by the vendor for each insert. The generated ED maps were normalized to the ED of water and used as the input for dose calculations without CT images. Dosimetry plans were generated on the phantom for two different field sizes (10 × 10 cm2 and 3 × 3 cm2) at gantry angles of 0 and 90 degrees using a 6 MV Monte Carlo engine. The dose distributions were compared between the conventional HU to ED calibration approach with CT images and the direct calculation using the calculated ED map. The results showed that compared to the conventional HU to ED map, the ED map generated from spectral CT had a relative ED that was about 0.02 lower and was more uniform, with smaller standard deviations. The ED map was closer to the nominal value in low-density regions, while the HU converted ED map was closer to the nominal value in high-density regions. The dose distributions between the two ED approaches were almost identical, with a maximum deviation of around 1% for both field sizes at deeper depths. In conclusion, a dual layer detector scanner can provide an accurate estimation of ED maps. We showed that the dose calculation using the generated ED map is highly accurate using a phantom. This method provides an alternative strategy for dose calculation that eliminates the need for HU to ED calibration and enables the use of the ED map directly.