68-6
32/104

Leaves passing (%)Absolute median error (mm)Maximum error (mm)596PF-like images using Pylinac. Table 5 shows the passing rates with a leaf tolerance value of 0.5 mm and the maximum error and average error of leaf positions. All the leaf positions were within a toler-ance of 0.5 mm at all gantry angles. The maximum error was 0.11 mm in the BPF plan with a gantry rotation of 360 degrees. In the plan with the artifi-cial error for MLC leaf positions, the passing rate was 95.2 % and the maximum error was 0.97 mm. The maximum artificially added leaf position error was 1.0 mm. Consequently, use of the PF-like image from the BPF plan can accurately detect leaf positioning errors. The PF test, as MLC QA, can be performed with the BPF plan.DiscussionThe JTT is effective to reduce the dose from radiation leakage transmitted by MLCs. The trans-mitted dose can reach 1.5% of the primary dose20). The uncertainty of jaw positioning affects the dose distribution delivered to a patient. In a study that evaluated the impact of jaw positioning uncertainty in conformal plans, a jaw positioning error of 10 mm can lead to a 5.0% dose error21). To date, few studies have investigated the influence of jaw posi-tioning uncertainty in VMAT with JTT, although some studies reported that jaw positioning uncer-tainty affects dose distribution6, 7). Therefore, it is important in high-precision RT to control the accu-racy of jaw position. In this study, we developed and evaluated a method to verify jaw positions in JTT delivery with a novel MLC motion pattern─the BPF pattern. This method uses an EPID equipped with a linac and freely available image processing software. Specific devices are not otherwise needed. Furthermore, the MLC motion pattern of the BPF is similar to a conventional MLC QA pattern, termed the picket fence pattern. A PF-like image can easily be created from the acquired images using the BPF pattern and analyzed using the conventional MLC QA method. Therefore, the veri-Table 5 Results of the picket fence test from Pylinac with BPF plan0 degreesfication of jaw positions in JTT and MLC positions can be performed simultaneously. Our proposed method allows verification with low cost and low workload, is quick, and provides accurate jaw posi-tion QA in jaw-tracking delivery.Detected jaw positions in the BPF plan with a static gantry were in good agreement with the planned positions. The absolute differences between the planned and detected positions were 0.16 ± 0.19 mm for the X1 jaw and 0.11 ± 0.16 mm for the X2 jaw. This finding suggests that jaw position can be detected even in a region where only transmitted radiation exists and a jaw image is difficult to see because of the radiation from the MLC-defined aperture. In particular, the region of MLC transmis-sion is smallest (23 mm) in the segment numbers 4 and 5 because the jaw aperture here is smallest among all segments (Figure 5). Even in such a situation, our method can detect jaw positions because Otsu thresholding works effectively to enhance contrast.The average of the absolute differences between the detected and planned jaw positions in the BPF pattern with the artificial errors were 0.14 ± 0.10 mm (average ± 1 SD) for the X1 jaw and 0.13 ± 0.15 mm for the X2 jaw. The maximum absolute differences were 0.26 mm for the X1 jaw and 0.38 mm for the X2 jaw. The gantry angle dependence was not observed.From Tables 1 and 2, we estimated the jaw posi-tion can be detected with a precision within 0.5 mm, because the maximum absolute difference was 0.31 mm (X1, segment 3, gantry angle 90 degrees). The report of American Association of Physicists in Medicine Task Group 142 recommends that the positioning accuracy of an asymmetric jaw should be within 1.0 mm22). Our method can satisfy the crite-rion, because the maximum absolute average ± 1 SD was 0.62 mm (X1, segment 1, gantry rotation).A similar method to simultaneously detect jaw and MLC positions is the machine performance Gantry angle90 degrees100.000.010.06100.000.030.09270 degrees360-degrees rotation100.000.020.05100.000.020.11

元のページ  ../index.html#32

このブックを見る