Modern Light Field /Radiation Field QA

Historical Notes

The Film Era

In the old days (1990's), we took a ready-pack film (Kodak X-Omat V), taped it on the treatment table, punched the corners with a needle, and exposed the film. After developing, on each field edge we searched for the 50% intensity point using a simple densitometer which allowed measurement of optical density in a small circular area, and marked it. This determined the radiation field size. By connecting the punched corner points with straight lines, the light field was established. Now light field size and radiation field size could be compared.

Since 2004, KFJ hospital is filmless. This marks the end of the film era.

The EPID Era

Since 2005, we use the aS500 portal imager to do the check in a filmless way:

(click to enlarge)

For each setup, a sequence of two images is acquired as double-exposure: the first image is taken at the target field size (e.g., 16 x 16 cm), the other with the jaws opened by 1 cm at all sides. The first image is an integrated image with 50 MU, the other a single shot.

Varian software detects the field edge of the first ("planned") image and projects the edge contour on the second ("open") image.

The object in the beam is a phantom plate with integrated steel BB's. If the jaws are set to match the light field on the BB's, the BB's mark the light field edge on the acquired double exposures. The central BB marks the light field center.

By comparing the BB's with the edge contour, the light field radiation field congruence can be checked.

While the method works well in general (also at Gantry 90°/270°) and gives an immediate qualitative visual result, quantitative evaluation is quite cumbersome. All distance measurements have to be done by hand, using the software tool.

Documentation is suboptimal too: we print out each image and write down the results. There is no database.

The EPID era is not over yet, because the method is still used for Gantry angles 90°/270°.

The post-imaging area

Here are some more pictures from the post-imaging era. For a photo of the STARCHECK^maxi, please visit the equipment page.

The FIELDCHECK test object.

The FIELDCHECK and the adapter frame which is needed for the STARCHECK^maxi.

During data acquisition, the real-time screen shows 4 profiles (6MV, 20 x 20 cm). The inplane and crossplane profiles are undistorted.

The diagonal profile is "distorted" by the FIELDCHECK objects in the beam (6MV, 20 x 20 cm). The pattern is used to calculate the LF size.

Same profile, different energy (15MV, 20 x 20 cm).

 

The light field (LF) of a medical linear accelerator is used to display the position of the radiation field (RF) on the patient skin. Although it is gradually losing it's importance due to IGRT, checking the LF/RF congruence is still mandatory in current QA protocols¹.

After following the standard procedures (see sidebar) for many years, we entered the post-imaging era in Feb 2011 with a new dosimetric approach that works without images. It uses the FIELDCHECK test object, the STARCHECK^maxi and MultiCheck software Version 3.4, all from PTW. Here is a typical MultiCheck screenshot:

The Principle

First the STARCHECK^maxi is set up on the treatment table and precisely aligned with the lasers and LF cross-hairs. Then the FIELDCHECK is placed on top of the maxi.² After the linac has moved the jaws to the target field size (10 x 10 cm or 20 x 20 cm), four movable blocks of the FIELDCHECK have to be adjusted by the user to match the LF edge.

When the field is measured (we use 100 MU @ 600 MU/min), the maxi acquires four profiles simultaneously, which are automatically evaluated by MultiCheck software: The inplane and crossplane profiles (which are acquired under homogeneous parts of the FIELDCHECK) are used to determine the radiation field size. The diagonal profiles on the other hand are "distorted" by metal wedges which are part of the FIELDCHECK's movable blocks. By analyzing these profiles, the software can calculate the LF size with high precision, see the poster by T.Perik et al.

Results

Starting in March 2011, we acquired data sets on two machines (Clinac 2300C/D, S/N 156 & 157) for two energies (6 MV, 15 MV), two field sizes (10 x 10 cm, 20 x 20 cm), and two source-detector distances (SDD = 100 cm and 150 cm³). For each field size, we made two successive shots, to check reproducibility.

Example 1: 6 MV, 10 x 10 cm, SDD 100 cm

Whereas all deviations regarding crossplane field size, inplane asymmetry and crossplane asymmetry were below 1 mm, the inplane field size showed some hysteresis. Depending on the direction of jaw movement to target field size, measured field sizes were either smaller (if the Y jaws closed in from larger fields) or larger (if Y jaws opened from smaller fields). In direction of movement, they "overshot" a little:

The amplitude of this hysteresis was in the order of 1.5 mm (see graph).

Each target field size was measured twice, and with either direction of jaw movement. This way a possible change in the Y-jaw hysteresis (e.g., due to increasing play in the readout potentiometer) could be safely detected.

The reason why hysteresis was not observed in the crossplane direction (X jaws) is probably due to the smaller distance of the Y jaws from focus: a small play in the readout potentiometer has a larger impact at isocenter distance for the Y jaws than for the X jaws.

By taking a closer look at field asymmetry, it became immediately clear that mainly the Y2 jaw of this specific linac showed the hysteresis. The reaction of Varian was that the hysteresis amplitude is within specs.

Example 2: 15 MV, 20 x 20 cm, SDD 150 cm

Although FIELDCHECK is not designed for SDDs other than 100 cm, it was no problem to measure at 150 cm. For setup, we recorded the CouchVrt reading at SDD 100 cm (e.g. 2.0 cm), added 50.0 cm, and moved the Couch down remotely to the new position, 52.0 cm⁴. Then we entered the treatment room and realigned the devices with LF cross-hairs. We were not interested in measuring lateral displacements due to non-vertical couch movement, but only the agreement between LF and RF.

In order to measure a 20 x 20 cm field at 150 cm, jaws have to be set to 13.3 x 13.3 cm at Isocenter. However, we defined this field size in MultiCheck as 20 x 20 cm.

Here are the results for inplane field size, crossplane field size, inplane asymmetry and crossplane asymmetry. The square indicates the reference measurement.

Discussion

After the devices have been properly set up, LF/RF measurements are quick, easy, accurate and reproducible. The only user-dependent factor is the judgement of the LF edge, when the user moves the four blocks of the FIELDCHECK. It requires some routine to get reproducible results at the 0.1 mm level. If different users work on the same dataset, larger variations can be expected for the LF.

We hope that some day a device will be available that is capable of "measuring" the LF and the RF at the same time, without being influenced by the "human factor".

Periodic measurements are always compared against a "reference" regarding field size and field asymmetry, for both the radiation field and the LF. This means that measurements (and the displayed deviations) are never "absolute", but have always to be seen "with respect to the baseline". Therefore, care must be taken when interpreting the results of a later routine check: any asymmetry or field size deviation which exists at the time of the baseline shot will be zero from now on.

However, displaying the reference measurement (here for 6 MV, 10 x 10 cm, Isocenter and 15 MV, 20 x 20 cm, SDD 150 cm) shows absolute values. This can, at least in principle (not within MultiCheck software), be used to get "absolute" values for all later measurements.

Two notes regarding the hysteresis effect:

• It is generally a good idea to leave some note in the comment field of each measurement which indicates the direction of jaw movement to target field size, e.g. "from+" or "from-". Otherwise it is difficult to interpret the results later on.
• The direction of jaw movement to target field size during the reference measurement will dominate deviations of later periodic measurements.

A limitation of the described method is that it currently works only for Gantry 0°. To perform such measurements at 90°/270°, a holder would have to be constructed with the ability to rotate the whole assembly (FIELDCHECK, STARCHECK^maxi and the connecting frame) to the upright position - without any mechanical play. This would be quite a challenge for the big and heavy maxi, and would be somewhat easier to accomplish with the smaller STARCHECK.

According to QA protocols, measurements at Gantry 90°/270° have to be performed. For this reason, we still take the 90°/270° shots with the aS500 port imager, as described in the sidebar.

 

¹The Austrian standard S 5290-1:2011-10-15 requires that LF/RF be checked twice a year.
²The adapter frame we use is not necessary if a smaller STARCHECK is used.
³The 150 cm are required by the Austrian QA standard.
⁴It has to be checked first that the vertical Couch readout is accurate at low couch positions, which are usually not used for patient treatments.