Session Item

The Elekta Unity linear accelerator (linac) is a high-field magnetic resonance (MR) linac allowing MR image-guided adaptive radiotherapy. The Unity linac has a build-in electronic portal imaging device (EPID) mainly for quality assurance (QA) and calibrations. The position of the central axis projection onto the EPID for gantry at zero degrees is essential for many QA procedures. The current study aims to create a simple phantom capable of measuring this zero position of the EPID, which can be used as an independent check of the vendor calibration.

Copper spheres (CSs), Æ4 mm, were pairwise placed vertically above each other. The CSs had a hole drilled centrally (Æ0.8 mm) such that they could be placed on a sewing thread, approximately 40 cm apart. The thread was passed through the hole and glued to fasten the position of the CS. Five strings, each with two CSs, were placed in a box to prevent the impact of air circulation. One line was placed close to the expected central beam, and the four others placed in the corners of a 6×6 cm square surrounding the centre string, Fig. 1(a). For varying gantry angles (±1º in step 0.1º), the CSs were irradiated with 20 MU while acquiring EPID images in both clockwise (CW) and counter-clockwise rotation (CCW). The projection of each pair of CS on the EPID will be on a line passing through the point of vertical beam direction, which for zero gantry position will be the zero position of the EPID. A MATLAB code was used to analyse the images for CS positions and perform cross-point calculations, Fig. 1(b).

The crossing point across and along the bore (U, V) on the EPID represents the vertical fan-line of the radiation beam at the actual gantry angle, Fig. 2. As expected, the crossing point in the direction across the bore (U) had a linear variation with gantry rotations U_fit =af+b, with a=114.8/º, b=510.5, and f being the gantry angle in degrees. Thus, the U cross point is sensitive to the definition of the zero gantry angle. The standard deviation of the crossing point of the four lines was approximately 1.3 (U) and 1.4 (V) pixels. The vendor calibration (Cal) is also shown in Fig. 2. A slight discrepancy between Cal and string phantom of 1.2 pixels (0.3 mm in iso-plane) and 3.4 pixels (0.7 mm in iso-plane) was observed in the U and V direction, respectively. The effect of CW and CCW rotations seems negligible.

A simple phantom able to detect the centre EPID position was developed. Only small differences between vendor calibration and phantom measurements were observed. The measured U direction depends on a correct definition of the zero position of the gantry. That uncertainty can be removed by using a single CS on the centre string in a Winston-Lutz test that can define the U position. However, only the string phantom can detect the EPID centre position in the longitudinal direction (V). Furthermore, the string phantom does not need precise alignment to correct measurements, making it easy and fast to use.