The effect of magnetic field strength on the response of Gafchromic EBT-3 film

The experimental setup (figure 1) involved irradiation of Gafchromic EBT-3 film in an electromagnet (GMW type 3474-140) using a ⁶⁰Co beam (Theratronics, Theratron 780c). The distance from the ⁶⁰Co source to the centre of the magnetic poles (measurement point) was 162 cm. An in-house polymethyl methacrylate (PMMA) phantom, 5 cm width and 26 cm height, was designed to fit inside the 5 cm gap between the two poles of the magnet (as shown in figure 1), which provide a maximum B-field strength of 2 T. The phantom consisted of two symmetrical PMMA plates, each at a water equivalent depth of 1 cm, placed in front of and behind the film, and located so that the central axis of the film is placed in the centre of the two magnetic poles. The edge of the film was orientated parallel to B-field, along the x-axis, and perpendicular to radiation beam, in direction of the z-axis (figure 1). The B-field uniformity in an area of 3  ×  3 cm² along the centre of the two magnetic poles (in z and y  direction) was found to be within  ±0.5 mT.

Zoom In Zoom Out Reset image size

Output measurements were made using alanine/EPR dosimetry, calibrated in terms of absorbed dose to water and traceable to the NPL primary standard, to determine the absorbed dose rate of the ⁶⁰Co beam at zero B-field. Nine alanine pellets, 2.5 mm height and 5 mm diameter, placed in a Farmer type (F-type) holder were used in a similar phantom to the one adopted for film dosimetry. Alanine was chosen due to non-standard condition of our experimental setup (such as field size and scatter conditions in the restricted space between the poles of the electromagnet). The phantom was built from a PMMA block, 2 cm  ×  5 cm  ×  26 cm, with a hole drilled into the insert to take an alanine F-type holder which could be fitted perpendicular to the direction of the radiation beam. The film and alanine phantoms were constructed to place the film and alanine reference points at the same measurement point.

Although the effect of a B-field is more pronounced in air cavities, it also modifies the distribution of dose in water. A study by O'Brien et al (2016), including Monte Carlo calculation of depth-dose curves at 0 T and 1.5 T with an ELEKTA MR-linac beam quality (7 MV), has shown that the 1.5 T field systematically decreases the dose by 0.5% at depths beyond ${{d}_{max}}$. The effect of a B-field on dose to water, needs to be taken into account and corrected for.

In the current work, ⁶⁰Co absorb dose to water was calculated in B-field strengths up to 2 T using the usercode cavity that forms part of the EGSnrc Monte Carlo system (Kawrakow et al 2011) (development version: GitHub: Aug 2017). The simulation used a water phantom to represent the PMMA phantom, with water equivalence defined by scaling depth inversely with electron density. Dose was scored in slabs of 0.05 cm thick and 1 cm wide up to a depth of 2 cm.

Films were prepared, scanned and analysed by using the method developed by Bouchard et al (2009). Briefly, 12 pieces of EBT3 films (3.5 cm  ×  3.5 cm), cut from one sheet of film, were used to generate a calibration curve. Each piece maintained the same orientation as the original short dimension of the film (landscape orientation) and placed parallel to the B-field during irradiation and parallel to scanning direction during digitisation. The film orientation and alignment was kept consistent throughout the measurements.

The radiation field filled the gap between the two magnetic poles and the source-to-surface distance was set at 161 cm. Films were irradiated at a depth of 1 cm (a sufficient amount of PMMA was used to provide charged particle equilibrium) and the maximum dose given at each B-field was approximately 825 cGy. A region of interest of 2.5 cm  ×  2.5 cm was analysed for each film in order to obtain the ${\rm netOD}$, where ${\rm netOD}$ is the difference between the optical density (OD) after and before irradiation. The film response curve is defined as ${\rm netOD}$ as a function of dose.

The films were digitised with an EPSON Expression 10000XL Pro flat-bed colour scanner in transmission mode. Coloured images were acquired with a spatial resolution of 150 dpi with 48 bit RGB and all scanner colour corrections turned off. A frame was used to position the films at an area of the scanner bed which could correct OD readings for scanner light non-uniformity (Saur and Frengen 2008). The orientation of all films was kept constant and aligned within  ±5° to avoid any effect due to polarised light. A 3.5 mm PMMA sheet was placed on top of the films during digitisation in order to position the films flat on the scanner bed. The scanner warming-up effect was diminished by using ten repeated scans. Films were scanned before, to account for background correction, and 48 h after irradiation in order to allow the film optical density to stabilise (Cheung et al 2005).

In a measurement of absorbed dose, the additional uncertainty⁵ introduced by using film arises from two sources: (i) the variations in film response from pixel to pixel within each ROI, which was determined using the method of Bouchard et al (2009), and (ii) the variations in response between ROIs on different pieces of film (in particular, the film used for measurement, compared with the films used for calibration), which was determined from a statistical analysis of repeated film measurements with 40 degrees of freedom.

Figure 2(a) shows the MC calculated ⁶⁰Co depth-doses for 0 T and 1.5 T B-field strengths and the ratio, dose from B-field of 1.5 T ($D_{{\rm w},{\rm Q}}^{{{{\rm B}}_{1.5}}}$) divided by the dose in zero B-field (${{D}_{{\rm w},{\rm Q}}}$), on the right y -axis. It can be observed that for a B-field of 1.5 T the surface dose is enhanced by 13% and the ${{d}_{max}}$ position is shifted closer to the surface by 0.05 cm. No effect was observed on the dose of ${{d}_{max}}$ between the two B-field strengths. Figure 2(b) shows the ratios of calculated dose in water with and without B-field as a function of a B-field strength. The ratios were determined from the average ratios between the depths of 0.5 cm and 1.7 cm and it was found to change by ‒0.08% per T when a linear model is fitted. Based on this change, the dose values of the EBT-3 calibration curve at each B-field strength were adjusted to account for the effect of the B-field on dose to water. The electron return effect could also be visible after a depth of approximately 1.7 cm.

Zoom In Zoom Out Reset image size

In this work we have analysed the red, green and blue channel. However, we present selected results in graphical form, but mostly for the red channel only, as this is the channel which is usually used for single channel film dosimetry.

The EBT-3 film response curves (expressed in terms of ${\rm netOD}$ as a function of dose) for the red channel for all B-field strengths considered in this study are shown in figure 3. The maximum spread between all the B-fields for the red channel is also presented on the right y -axis of the same figure. For dose values greater than 75 cGy the maximum spread is 2.6% for the red channel, 3.5% for the green channel and 4.2% for the blue channel.

Zoom In Zoom Out Reset image size

Table 1 shows the average absolute difference (in percent) of all dose levels between the response curves irradiated with and without a B-field. The difference is presented for each B-field and each colour channel.

We define $\Delta {\rm netO}{{{\rm D}}_{B}}$ to be the difference between the ${\rm netOD}$ with and without a B-field:

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \Delta {\rm netO}{{{\rm D}}_{B}}={\rm netO}{{{\rm D}}_{B\ne 0}}-{\rm netO}{{{\rm D}}_{B=0}}.\nonumber \end{align} \tag{ 1 }$$

Figure 4 shows $\Delta {\rm netO}{{{\rm D}}_{B}}$ as a function of dose, for each B-field and for the red channel only. The error bars denote the standard uncertainty in ${\rm netOD}$. The maximum $\Delta {\rm netO}{{{\rm D}}_{B}}$ was found to be 0.006 for the red and green channels and  −0.003 for the blue channel. Figure 4 shows that $\Delta {\rm netO}{{{\rm D}}_{B}}$ does not have a strong dependence on dose. Similar results were observed for the green and blue channels.

Zoom In Zoom Out Reset image size

It appears that $\Delta {\rm netO}{{{\rm D}}_{B}}$ does depend on the B-field strength, being typically negative at 0.5 T and positive at 1 T and above. This effect is shown more clearly in figure 5, which depicts the average $\Delta {\rm netO}{{{\rm D}}_{{B} }}$ over all dose levels, ${{\left\langle \Delta {\rm netO}{{{\rm D}}_{B}} \right\rangle }_{{{D}_{{\rm all}}}}}$, as a function of B-field strength, for each channel.

Zoom In Zoom Out Reset image size

The effect of the B-field on dose measured using EBT-3 films is shown in figure 6. The calibration curve, based on films irradiated at zero B-field, was used to obtain absorbed dose from films irradiated at non-zero B-field, which would be valid if the B-field has no effect on film response. Figure 6 shows the ratios, dose obtained from non-zero B-field films (using red channel) divided by the dose actually delivered. The delivered dose value takes account of the small effect of the B-field on dose to water, which was calculated by MC. The extent to which these ratios are significantly different from unity indicates the effect of the B-field on the film response. For all B-field strengths the ratios decrease with dose, up to 225 cGy. Such behaviour would be accounted for by the inability of a fourth order polynomial to adequately represent film response at low dose. For values above 225 cGy, the ratios as a function of dose do not follow a simple trend and are somewhat scattered. Indeed, all ratios below 2 T are unity within the uncertainties.

Zoom In Zoom Out Reset image size

Taking averages over all dose values above 75 cGy, the mean ratios are 0.994  ±  0.024 (0.5 T), 1.011  ±  0.023 (1 T), 1.012  ±  0.028 (1.5 T) and 1.024  ±  0.026 (2 T).

The development of dosimetry in MRIgRT is still at a fairly early stage and it is essential that the performance of detectors used is well characterised. In the current literature on film dosimetry in B-fields, there remains some apparent inconsistency in the results reported from different studies. With the aim of resolving any consequent confusion we compare, where possible, the results reported here with those already published.

Roed et al (2017) examined the effect of a B-field (0 T and 1.5 T) on the response of EBT-3 film signal exposed in a ⁶⁰Co radiation beam. Their experiment involved two different orientations of the reference film edge with respect to (a) the B-field and (b) to scanning direction: either parallel or perpendicular in both cases. In all possible combinations, when film analysed using red channel, they found an under-response (ranging from approximetly 1% to 3%) on all examined dose levels with B-field strength of 1.5 T compared to zero B-field dose. However, they also state that when the film is left in the B-field for a duration of (a) 6 min and 30 min the dose is over responding by less than 1% and (b) 7 min and 10 min the dose is under responding by less than 1%. The relationship between the dose change and the time of the film left in B-field they do not follow a simple pattern. In our study, the irradiation time of the films was ranging between 4 min to 15 min and the dose was found to over respond by an average of 1.2% at 1.5 T.

Reynoso et al (2016) used a ⁶⁰Co 0.35 T MRIdian system to irradiate EBT-2 films. Their results indicate that the ${\rm netOD}$ of the red, green and blue channel is decreased by an average of 8.7%, 8% and 4.3%, respectively. This study concludes that B-fields affect crystal orientation and polymerization during irradiation, judging from films scanned with an electron microscope. Although that study evaluated a different type of film, they specifically state that similar behaviour can be expected for EBT-3 because of the identical composition of the active layer. In the present work, the effect of a B-field on the netOD response of EBT-3 was found to be small. In the Reynoso et al (2016) study they show an average under response of netOD of approximately 7% for data up to 8 Gy (figure 2(b) in Reynoso et al (2016)). The reason for the difference between the results of these two studies is unclear.

A publication by Delfs et al (2018) assessed the effect of the B-field on EBT-3 using a 6 MV conventional linac and irradiating films in an electromagnet at two different magnetic strengths (0.35 T and 1.42 T). They report a decrease in the OD of up to 2.5% with an average, over all channels and over a range of dose values, of  −0.8%. This effect was described by the authors as 2.1% increase in the dose values needed to produce a given OD at a B-field of 1.42 T. Our study found that there is indeed a change on film sensitivity however in the opposite direction of Delfs findings: the net OD is increased by  +0.8% (red channel, 1.5 T). The dose required to produce a given optical density is smaller by 1.2%. Part of this difference might be explained due to energy dependence of EBT-3 film.

A recent study by Barten et al (2018) investigated the suitability of EBT-3 film for QA use in a 0.35 T ⁶⁰Co MRIdian system. Their measurements involved irradiation of EBT-3 film with and without MR imaging and at different angles with respect to B-field. They also performed equivalent measurements at 0 T, but this time using a 6 MV conventional linac. Comparing calibration curves generated in ⁶⁰Co, with B-field, and in 6 MV, without B-field, they found a maximum difference of 0.9%. In our study the difference on dose measured using EBT-3 film with and without B-field in a ⁶⁰Co beam and at 0.5 T (the closest strength to 0.35 T) was found to be 0.7%, which agrees with the findings from Barten et al (2018).

The observed differences between the current study and the study by Roed et al (2017) show to be consistent within film uncertainties. Due to the nature of the film technique, any deviation can be explained because of the film perturbations influencing measurements uncertainties. However, this cannot explain the difference with the findings from Reynoso et al (2016) as it exceeds uncertainty margins. Although different type of EBT film generations were used, we anticipated consistency of published data and that EBT-2 to have a similar behaviour to EBT-3 under the strength of B-field.

It is essential to take measurement uncertainty into account when assessing the degree of agreement between two investigations. The field strength used by Delfs et al (2018) was 1.42 T. Assuming that it is fair to compare their result with our result for 1.5 T, and assuming that no correction is made for the possible effect of B-field on a film measurement of absorbed dose, then Delfs et al (2018) would estimate the uncorrected error as  +2.1%, and we would estimate the uncorrected error as  −1.2%. We quantify our standard uncertainty on this error as 1.4%, while Delfs et al (2018) merely assert that their 2.1% value 'exceeds the uncertainty margin of the experiment' but without quantifying the uncertainty on dose measured by film in their paper. We would note, however, that the difference between our results could only be significant at the level of two sigma if their standard uncertainty on dose is smaller than 0.8%, which is not plausible.

A good agreement was found between our study and the study of Barten et al (2018). Each of these studies (Barten et al 2018, Delfs et al 2018), however, used different beam quality, or a combination of beam qualities, to examine the effect of B-field on EBT-3 film response, compared to our work.