Variability of HU–Density Calibration Curves across Ethos Hypersight CBCT and CT Imaging Systems
Abstract
Purpose
Accurate Hounsfield Unit (HU) to density calibration curves are essential for reliable dose calculations in adaptive radiotherapy. This study quantifies variations in HU-mass density and HU-electron density relationships across imaging protocols, and Ethos Hypersight CBCT and CT units.
Methods
A Sun Nuclear Electron Density phantom with tissue-equivalent inserts (lung, adipose, breast, muscle, liver, and bone) was scanned on two GE Discovery CT Scanners and three Ethos machines across two institutions using a range of clinically-relevant imaging protocols and energies. Tissue-equivalent region of interest measurements were compared between the two Ethos linacs at one institution to assess inter-unit variability. The calibration curves derived for these units was then compared against the calibration curve of the Ethos unit at another institution and two CT scanners with pointwise density differences.
Results
The two Ethos machines had five 125 kV imaging protocols and two 140 kV imaging protocols in clinical use. The largest variations inter-unit variations were observed for bone-equivalent material, while soft tissue-equivalent regions showed smaller differences. Reduced variability between units was observed for higher energy (140 kV) protocols. Up to 1000 HU, the density differences between Ethos and CT derived calibration curves were within 0.05 g/cm3 and 0.04 electrons/cm3. The root-mean-square difference was ≤ 0.03 g/cm3 for HU-mass density curves and ≤ 0.02 electrons/cm3 for the HU-electron density curves, indicating reasonable similarity between the curves.
Conclusion
HU–density calibration curves were generally consistent across Ethos Hypersight CBCT and CT systems. Some inter-unit variations were observed when different imaging protocols were used, particularly for high-density materials. While a single calibration curve is appropriate for use across multiple systems, it is important to be aware of potential machine, energy, and protocol dependent variations.