Documentation for the project in PHYS291 - data handling in physics

Impact of proton spot size on dose and Linear Energy Transfer (LET) distributions using Monte Carlo simulations



Introduction

In proton therapy, there are two ways to deliver the proton beam to the target (tumour): passive scattering and active scanning. In the latter, a narrow beam guided by magnets is delivered in the target, pixel-by-pixel, in successive layers. Each pixel represents a spot. The size of these spots are determined by the full width at half maximum (FWHM) of the primary proton beam. How to determine the size of the spots is dependent on the the programme used to make the simulations. One can either use programmes that vary the spot sizes directly, or programmes that vary the spacing between the spots. The latter is the case for the simulation programme used for this project. As the relationship between spot size and spot spacing is linear, changing the spot spacing is equivalent to changing the spot size - widening the spacing between the spots increases the spot size, and vice versa. Will changing the spot size have an impact on dose and LET distributions in the target volume?

According to the theory, changing the spot spacing - and thereby the spot size - should not have a significant impact on the dose and LET distributions. By plotting this distribution for three different spot spacings in two plots - one for the LET distribution and the other for the dose distribution - one may be able to see if this is in agreement with the theory.

Active scanning proton beam delivery. The pixels represent the spots.



Procedure

Using Monte Carlo (MC) simulations, we will investigate the impact of proton beam spot size on the dose and LET distribution in a water phantom. For the LET distribution, we will look at the dose averaged LET, called LETd, which is an average value of the LET spectrum.

The project was done according to the following steps:

1. Create simple dose plans for a water phantom in the MatRad software, and vary the proton spot spacing. The spacings will be 3mm, 5mm and 7mm.
2. Score dose, LET and LET-squared for all three setups using FLUKA Monte Carlo simulations based on the plan from MatRad, and export the data.
3. Write a C++ script to get LETd from the MC output files for LET and LET-squared.
4. Write a second C++ script which reads the MC output files for the dose and the calculated LETd output files, and extracts/calculates relevant parameters and provides data in a format suitable for further handling/visualisation in ROOT.
5. Plot/visualise the results using ROOT.
6. Analyse the results.

The scripts used for the project:
Calculating LETd - using the MC output files for LET and LET-squared to calculate LETd in a new data file.
Output dose and LETd - using the MC output files for dose and the calculated LETd files to create text files for dose and LETd for all three spot spacings.
Plotting LETd - using the created text file for LETd to plot the LETd distribution in ROOT.
Plotting dose - using the created text file for dose to plot the dose distribution in ROOT.
The zip file for all four scripts is found here.

Results

The results of this project are found in the figures below.

The plot for the LETd distribution:

Plot 1: Calculated LETd in the water phantom.


The plot for the dose distribution:

Plot 2: Dose distribution in the water phantom.


Here, one can see that by changing the spot spacing from 3mm to 5mm and then to 7mm, the impact this has on both the LETd and dose distribution is not very significant. In other words, the results seem to be in good agreement with the theory.

Although the value of the uncertainties shown by the errorbars are roughly the same for each of the spot spacings, the large scale of the uncertainties shown in the plot for dose distribution should be looked at. Due to the scaling of the y-axis for each of the two plots, it is difficult to see the exact values of the deviations. By looking at the text files used to generate the two plots, it is evident that the uncertainties are much lager than one would expect for the dose (~70%), especially when compared to the uncertainties for the LETd (~1.4%). The most likely explanation for this is that something has gone wrong in the scoring of the dose in the FLUKA Monte Carlo simulations. The dose distribution in terms of errorbars from the FLUKA Monte Carlo simulations are shown below, to illustrate that the large uncertainties are already in place in the scoring of the dose.

Dose distribution in the water phantom from FLUKA, 3mm spot spacing.
Dose distribution in the water phantom from FLUKA, 5mm spot spacing.
Dose distribution in the water phantom from FLUKA, 7mm spot spacing.


Discussion

The main goal of this project was to see whether changing the spot sizes would have an impact on the dose and LETd distribution in a target volume. This was done by creating simple dose plans for a water phantom, using the output files created by the FLUKA Monte Carlo simulations for further calculation and rewriting the files in a format suitable for plotting in ROOT. The resulting plots for dose and LETd distribution show an agreement with the theory that varying the spot size would have minimal impact on these distributions. The large uncertainty values for the dose distribution are a different matter. Further work for this project could therefore include finding where the uncertainty values from the scoring went much higher than one would expect, and make the appropriate corrections in FLUKA to get better uncertainty values than the ones shown here.

References

The information about active scanning, as well as the picture displaying active scanning in the introduction was taken from a lecture slide on proton therapy held by Kristian Ytre-Hauge during the course PHYS213 at UiB in the autumn of 2016 ("Beam delivery", slide 47-70). The rest of the background information for this project was given through conversations with Kristian Ytre-Hauge and Tordis Dahle.