On the prediction of X-ray dose deviations and the influence of CT scan protocols

Patients undergoing computerized tomography (CT) scans for tumor localization and treatment planning are frequently scanned using pre-set customized exposure protocols for optimal imaging of different anatomical sites. The question arises if these scanning protocols will produce a deviation in the Hounsfield number for a given tissue that can afterwards be used to predict the resulting dose calculation deviation due to this. The question is also if the deviation in the Hounsfield number of a tissue is large enough to affect dose calculation clinically significant.

A study was devised in which a RMI phantom was scanned with five different scanning protocols and two CT beam energies at 120 and 135 kV. To assess the effect of insert configuration, Hounsfield number measurements were repeated for high density RMI inserts in the center and outer rings in the phantom. For each material insert the standard deviation of the Hounsfield number was calculated. To assist in dose prediction a series of DOSXYZnrc Monte Carlo calculations were carried out for beam qualities between 6 and 16 MV for a range of Hounsfield numbers calculated for bone and water. This provided information on how the depth dose varied as a function of Hounsfield number variation. Lastly, a series of treatment plans were setup for absorbed dose calculation using the RMI insert electron densities vs Hounsfield relations measured above. The absorbed dose of corresponding plans with the largest Hounsfield number variation were subtracted to find the dose discrepancies.

It was found that the dose discrepancies in tissue types could be indicated by the deviation of the Hounsfield number due to different scanning protocols. The calculated dose difference were in all cases within 3%.