Logo of the Physikalisch-Technische Bundesanstalt

Development and benchmarking of an adaptive Monte Carlo model for CT scanners

28.07.2022

Examinations by means of computed tomography (CT) are responsible for about 65 % of artificial radiation exposure in Germany. The aim of personalized CT dosimetry is to individually measure the radiation dose transferred to a particular patient by the radiation field of the CT scanner. High‑precision Monte Carlo (MC) computer simulations are of utmost importance for achieving new developments in personalized CT dosimetry.

Based on a previously developed method for the experimental characterization of radiation fields of arbitrary CT scanners (CT source), a customizable MCCT source has now been integrated into the MC software EGSnrc and experimentally validated.

Surveys by the Federal Office for Radiation Protection (2014) show that about 65 % of the exposure of the population to artificial radiation is caused by medical imaging using CT. CT thus causes the largest share of artificial radiation exposure within Germany as a whole. The accurate recording of the radiation dose transferred to humans by CT radiation (CT dosimetry) is therefore essential. The aim of personalized CT dosimetry is to individually measure the radiation dose transferred to a particular patient by the radiation field of the CT scanner.

High‑precision MC computer simulations are of utmost importance for achieving new developments in personalized CT dosimetry. They are used, for example, to generate synthetic training data for neural networks for rapid dose calculation. Detailed modeling of arbitrary CT scanners is essential for the MC computer simulations required for this purpose. However, this poses particular challenges, since detailed descriptions of the components used in the CT scanners (for example, the bow‑tie filter), are often unknown and thus difficult to model in the computer.

Based on a previously developed method for the experimental characterization of radiation fields of any CT scanner (CT source), a customizable MCCT source has now been integrated into the MC software EGSnrc. This allows scanner‑specific calculations of the effective dose based on patient or phantom data. To validate the method, one such MCCT source was adapted to an Optima CT 660 CT scanner (GE Healthcare, US). A series of measurements was then conducted on the CT scanner. These measurements were replicated using the adapted MCCT source and the measurement results then compared with the calculated data.

Figure 1 shows the measured and simulated attenuation of the radiation field through the bow‑tie filter as a function of the angle to the beam axis. To evaluate the photon spectrum of the radiation field an aluminum attenuation curve was also determined (see Figure 2). Both figures show that the matched MCCT source reproduces the measurements on the real CT with deviations of just a few percent. In addition, dosimetric measurements and simulations were performed and compared within a CTDI phantom.

diagram attenuation of the radiation field
Figure 1: Measured and simulated attenuation of the radiation field through the bow-tie filter, as a function of the angle to the beam axis.
diagram simulated and measured aluminium attenuation curve
Figure 2: Comparison of the simulated and measured aluminum attenuation curve.

The deviations between the measured values and the values calculated by the MC method are below 8 % for all investigations.

As such, the developed method allows CT dose calculations based on a customized MC model that can be adapted to any CT scanner as well as to patient‑specific information.

This method is thus suitable for the automatic generation of synthetic data. This is an important step towards training neural networks for fast dose calculations.

Contact

Opens local program for sending emailMarie-Luise Kuhlmann, Department 6. 2, Working Group 6. 25