A quantitative method to the analysis of MLC leaf position and speed based on EPID and EBT3 film for dynamic IMRT treatment with different types of MLC

Abstract A quantitative method based on the electronic portal imaging system (EPID) and film was developed for MLC position and speed testing; this method was used for three MLC types (Millennium, MLCi, and Agility MLC). To determine the leaf position, a picket fence designed by the dynamic (DMLC) model was used. The full‐width half‐maximum (FWHM) values of each gap measured by EPID and EBT3 were converted to the gap width using the FWHM versus nominal gap width relationship. The algorithm developed for the picket fence analysis was able to quantify the gap width, the distance between gaps, and each individual leaf position. To determine the leaf speed, a 0.5 × 20 cm2 MLC‐defined sliding gap was applied across a 14 × 20 cm2 symmetry field. The linacs ran at a fixed‐dose rate. The use of different monitor units (MUs) for this test led to different leaf speeds. The effect of leaf transmission was considered in a speed accuracy analysis. The difference between the EPID and film results for the MLC position is less than 0.1 mm. For the three MLC types, twice the standard deviation (2 SD) is provided; 0.2, 0.4, and 0.4 mm for gap widths of three MLC types, and 0.1, 0.2, and 0.2 mm for distances between gaps. The individual leaf positions deviate from the preset positions within 0.1 mm. The variations in the speed profiles for the EPID and EBT3 results are consistent, but the EPID results are slightly better than the film results. Different speeds were measured for each MLC type. For all three MLC types, speed errors increase with increasing speed. The analysis speeds deviate from the preset speeds within approximately 0.01 cm s−1. This quantitative analysis of MLC position and speed provides an intuitive evaluation for MLC quality assurance (QA).


| INTRODUCTION
DMLC technology has been widely used in intensity modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT), given its better tumor dose conformity and reduced radiation to the organs at risk. 1,2 In order to gain the actual clinical advantage from treatment, it must be ensured that the DMLC technology is performed accurately according to the treatment planning parameters.
The precision of DMLC technology depends on the accuracy of the leaf position and speed. 3  in leaf speed can result in increased beam holds or gap width errors. 6 The dosimetric effects of leaf speed errors are reported by Daniel et al. 7 and Huang et al. 8 An acceptance quantitative criterion has been proposed for MLC leaf position and speed in AAPM Task Group report 142. 6 Thus, it is necessary to make a quantitative assessment for leaf position and speed in routine MLC QA.
The picket fence is the most commonly used test for the MLC leaf position. Chui et al. 9 were the first to design a picket fence with DMLC, and the leaf position errors were evaluated by visual inspection. Chang et al. 10 designed a picket fence with DMLC, and images were acquired by the electronic portal imaging system (EPID) and film for digital analysis. The full-width half-maximum (FWHM) values of the gaps and the distance between gaps (interpeak distance) were used to evaluate the consistency of the leaf position, and the standard deviation was used as an evaluation criterion. Given the difference between the FWHM values of the gap and gap width, the system leaf position errors cannot be directly analyzed in a quantitative manner using the FWHM.
A sliding-window field can be used to determine the MLC leaf speed. Ling et al. 11 evaluated the MLC leaf speed control during Rapi-dArc. Their field contained multiple DMLC subfields. The exposure dose for each subfield was the same, and the dose rates for each subfield were differenced to combine different leaf speeds. Open-field profiles of the same size were used as evaluation criteria to determine the dose deviation of the sliding field caused by leaf speed errors. Currently, the tools used for MLC position and speed tests are primarily based on EPID and film methods. [12][13][14][15][16] As a quick and convenient measurement tool, EPID has been widely used in MLC QA work, 17,18 and the film approach is also widely used in MLC QA work as a high-resolution tool.
In this study, a single quantitative evaluation method for determining the leaf position and speed was developed for the above three MLC types. The EPID and EBT3 film approaches were used in this QA work to quantify the leaf position and speed errors. In addition, the results obtained from these two tools were compared.

2.A | Linear accelerator and MLC
The linacs used in this study included Trilogy (Varian, Palo Alto, CA, USA) with 120 Millennium leaves, Synergy (Elekta, Crawley, UK) with 80 MLCi leaves and Versa HD (Elekta, Crawley, UK) with 160 Agility leaves. The 6-MV photon mode was used for all irradiation. Before the measurement, the machine performance for all linaces has been checked according to the AAPM Task Group report 142. 6 The Millennium MLC consists of two banks of 60 leaves: the central 40 leaves of each bank are 0.5 cm in width (at the isocenter plane) and the outer 20 leaves are 1.0 cm in width. The minimum gap width formed by the Millennium MLC is 0.5 mm, and the maximum leaf speed is 3.0 cm s À1 . The MLCi MLC consists of two banks of 40 leaves, and the leaf width is 1.0 cm at the isocenter plane.
The minimum gap width formed by the MLCi MLC is 5 mm, and the maximum leaf speed is 2.0 cm s À1 . The Agility MLC consists of two banks of 80 leaves, and the leaf width is 0.5 cm at the isocenter plane. The minimum gap width formed by the Agility MLC is 4 mm, and the maximum leaf speed is 3.5 cm s À1 .

2.B | EPID and film
The Trilogy system includes the aS1000 EPID (Varian, Palo Alto, CA, USA) for megavoltage image acquisition. The source to detector distance (SDD) is 140 cm. The sensitive area of aS1000 is 40 9 30 cm 2 , containing 1024 9 768 pixels, and the pixel pitch is 0.392 9 0.392 mm 2 . Image acquisition was performed in integrated mode, and offset correction, gain correction, and pixel correction were performed for each image. The EPID image data were then back-projected to the isocenter plane for analysis. A linear relationship between the EPID pixel value and dose value was verified in previous studies of our group. 19,20 Therefore, to simplify the procedure, pixel values were used to analyze the leaf position and speed.   In this work, the field size was 14 9 24 cm 2 and the gantry was set at zero degrees. The collimators of Synergy and Versa HD systems were set at zero degrees, and the Trilogy system was set at 90°to include as many MLC leaves as possible. Eight pickets were formed by a sliding gap stopping at every 2 cm. The total beam weight is 1 To verify the accuracy of the MLC position analysis presented in this work, a deviation of AE 1 mm for the distance between the gaps, of AE 1 mm for the gap width, and of 1 mm for the position for only left or right leaves were introduced in gap 2, gap 3, and gap 4, respectively, as shown in Fig. 1, using the EPID image of the Millennium MLC as an example. Each measurement was repeated six times to reduce any uncertainty.

2.C.2 | MLC speed test
In this work, a 0.5 9 20 cm 2 MLC-defined gap was moved across a 14 9 20 cm 2 symmetric field at a constant speed. The linacs were run at a fixed-dose rate (600 MU/min theoretical dose rate), and both the collimators and gantry were set at zero degrees. We then sought to verify the accuracy of the MLC speed analysis in this work. For a sliding field with a normal speed of 1 cm s À1 , the monitor unit deviations (3, 2, 1, À1, À2, and À3 MU) were introduced into the 3-cm length of the center of the sliding field. The

2.D | Data analysis
To reduce the effect of interleaf leakage, 10 data regarding the leaf center position from each leaf pair were used for analysis. In addition, to reduce the effect of noise, 23 the middle nine profiles were averaged in the direction perpendicular to the leaf motion, and the average values were used for analysis.

2.D.1 | MLC position test
The six gaps (gaps 1-6 in Fig. 1) in the picket fence were used as analysis data for the MLC position. A QA of the MLC positions was performed based on the results for the distance between the gaps and the gap width. If leaf position errors were found, determining the individual leaf position for each leaf pair was necessary.

Calibration of gap width
The picket fence images from Trilogy, Synergy, and Versa HD each contained a 264 (44 leaf pairs 9 6 pickets), 144 (24 leaf pairs 9 6 pickets), and 288 (48 leaf pairs 9 6 pickets) FWHM values, respectively. The average FWHM over all leaf pairs was used to calibrate the gap width. According to the FWHM versus nominal gap width relationship (Fig. 2), an appropriate calibration function The measurement gap width (mgap w ) for the position test field can be calculated from the calibration function.

Determination of individual leaf position
The position errors can be easily detected by measuring the gap width and distance between gaps for each leaf pair. The next study will determine the position for each individual leaf in the testing field.
Using one leaf pair as an example, the individual leaf position on the left and right of each gap can determined by formulas (1) and (2).
where P L and P R are the leaf positions on the left and right of the gap, respectively, and P is the peak position of the gap.

2.D.2 | MLC speed test
One leaf pair was taken as an example to analyze the leaf speed.
For a fixed-dose rate, the dose contribution at the j-th position was determined as follows: where D j is the irradiation dose rate at the j-th position; is the transmission dose rate at the j-th position; where is the transmission dose; and   Table 5. The average of six measurements is presented with AE 2 SD. The average difference between the nominal and measured values is less than 0.1 mm, indicating that the algorithm developed for the picket fence analysis based on the EPID and film methods can accurately detect the MLC leaf position, including the gap width, the distance between gaps, and each individual leaf position. A verification of the consistency between the sliding and openfield profile measured by EPID and EBT3 film is presented in Fig. 3. Therefore, the open-field profile can be used as a standard to determine the dose-rate ratio w j , and the measurement data ranging from À5 to 5 cm were used for leaf speed analysis.

3.B | Speed test
The K 0 value in eq.  Table 7.
T A B L E 4 Leaf position results for the three MLC types measured by the EPID and EBT3 film methods, including the gap width and the distance between gaps. The average data over all leaf pairs AE 2 SD are provided.

| 111
The sliding profiles for the Trilogy MLC at the normal leaf speed and introduced leaf speed difference measured by EPID and EBT3 film are presented in Fig. 6. The profiles for Synergy and Versa HD are similar to those for Trilogy. As noted in Fig. 6 However, for the other measurement, the transmission dose is greater than 10%, and an accurate speed analysis must consider the influence of transmission. Detailed statistical results are presented in Table 8.

| DISCUSSION
In this study, it was found that the method can quantify the MLC leaf position and speed accurately by using EPID and EBT3 film.
Three MLC types were investigated in this study. It was found that the position and speed QA for all of these MLC types can be achieved using this method.
Compared with the 1-mm gap width utilized for the picket fence test in many studies, 9,10,18 the gap widths in this work (5 mm for Varian and 10 mm for Elekta) were significantly greater.  The larger gap width in our work can be calibrated by a linear function (as shown in Fig. 2), which can be fitted with fewer data points compared with those for a nonlinear calibration; thus, the calibration is simple and quick. The larger gap width may reduce the ability of visual inspection (in Fig. 1, the 0.5 mm errors can still be clearly observed), but the sensitivity of the data analysis is not affected.
Using the FWHM to calibrate gap width, the advantage is that For the evaluation of leaf speed, the effect of transmission has been considered in this study. If the effect of transmission not considered, the simple visual inspection for leaf speed may not be affected (in Fig. 6), but the quantitative results have significant differences with the introduced speed errors. The difference was increased with increasing speed errors. Therefore, an accurate quantitative analysis of leaf speed must consider the impact of transmission.
Although the analysis results in this work have demonstrated that both EPID and EBT3 film can be used for MLC speed QA, the lower signal-to-noise ratio is an issue for leaf speed QA in EBT3 film.
To improve the signal-to-noise ratio, each film in this study was repeatedly irradiated (more than three times), and the average data F I G . 6. The sliding profiles for Trilogy at the normal leaf speed and introduced leaf speed difference measured by EPID and EBT3 film.
T A B L E 7 The statistical results of leaf speed for the three MLC types measured by EPID and EBT3 film, using one leaf pair as an example. The averages of the speed profiles are presented with AE 2 SD. were used for analysis. Both the EPID and EBT3 results indicate that the leaf speed errors increase with increasing leaf speed. The same results are evident in the study by Rowshanfarzad et al. 18 Due to the limitation from size or SDD of EPID and film, the iView GT EPID and EBT3 film need two measurements to evaluate all MLC leafs. The aS1000 EPID can evaluate all MLC leafs in a single measurement with the conditions of 90°collimator angle and SDD100 cm.

| CONCLUSION
This study provides a quantitative analysis method for MLC position and speed QA based on the EPID and EBT3 film approaches. The method can be applied to multiple MLC types and ensures safe and reliable use of the DMLC IMRT.

CONF LICT OF I NTEREST
The authors declare no conflicts of interest.