Institute of Physics and Engineering in Medicine Phys. Med. Biol. 60 (2015) 7101–7125
Physics in Medicine & Biology doi:10.1088/0031-9155/60/18/7101
Modulation index for VMAT considering both mechanical and dose calculation uncertainties Jong Min Park1,2,3,4, So-Yeon Park1,2,3,5 and Hyoungnyoun Kim6 1
Department of Radiation Oncology, Seoul National University Hospital, Seoul, 110–744, Korea 2 Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, 110–744, Korea 3 Biomedical Research Institute, Seoul National University Hospital, Seoul, 110–744, Korea 4 Center for Convergence Research on Robotics, Advanced Institutes of Convergence Technology, Suwon, 443-270, Korea 5 Interdiciplinary Program in Radiation Applied Life Science, Seoul National University College of Medicine, Seoul, 110–799, Korea 6 Graduate School of Information, Yonsei University, Seoul, 120–746, Korea E-mail:
[email protected] Received 15 May 2015, revised 28 July 2015 Accepted for publication 4 August 2015 Published 28 August 2015 Abstract
The aim of this study is to present a modulation index considering both mechanical and dose calculation uncertainties for volumetric modulated arc therapy (VMAT). As a modulation index considering only mechanical uncertainty of VMAT, MIt has been previously suggested. In this study, we developed a weighting factor which represents dose calculation uncertainty based on the aperture shapes of fluence maps at every control point of VMAT plans. In order to calculate the weighting factor, the thinning algorithm of image processing techniques was applied to measure field aperture irregularity. By combining this weighting factor with the previously suggested modulation index, MIt, comprehensive modulation index (MIc) was designed. To evaluate the performance of MIc, gamma passing rates, differences in mechanical parameters between plans and log files and differences in dosevolume parameters between plans and the plans reconstructed from log files were acquired with a total of 52 VMAT plans. Spearman’s correlation coefficients (rs) between the values of MIc and measures of VMAT delivery accuracy were calculated. The rs values of MIc ( f = 0.5) to global gamma passing rates with 2%/2 mm, 1%/2 mm and 2%/1 mm were −0.728,−0.847 0031-9155/15/187101+25$33.00 © 2015 Institute of Physics and Engineering in Medicine Printed in the UK
7101
J M Park et al
Phys. Med. Biol. 60 (2015) 7101
and −0.617, respectively ( p αfσMLC accel)∙WGA,i+ 1∙WMU,i+ 1 WAI,i means if the MLC speedi is larger than the value of fσMLC speed, or the MLC acceli is larger than the value of αfσMLC accel, then the value of that ith CP becomes 1 and is multiplied by WGA,i+ 1, WMU,i+ 1 and WAI,i. If the conditions are not met, the value of that ith CP becomes 0 and it is not counted. In other words, this is a weighted counting by WGA,i+ 1, WMU,i+ 1 and WAI,i. The sum was over all CPs in a plan. After that, for each MLC, we calculated individual MIc,n of nth MLC as follows.
∫0
k
Individual MI ( f )df (4) c, n =
where, k = 0.2, 0.5, 1 and 2 in this study. 7108
J M Park et al
Phys. Med. Biol. 60 (2015) 7101
The MIc was calculated as follows. 120
∑
MI individual MI c, n (5) c = n=1
The sum was over all MLCs. The values of MIc were always larger than the values of MIt due to the WAI,i term. As both mechanical and dose calculation uncertainty increase, the value of MIc increases. 2.2. Volumetric modulated arc therapy plans
For performance evaluation of MIc, a total of 52 VMAT plans from 2 institutions were selected in this study. The plans consisted of 22 prostate and 30 head and neck (H&N) VMAT plans. Forty VMAT plans (20 prostate and 20 H&N VMAT plans) were from institution 1 and 12 VMAT plans (2 prostate and 10 H&N VMAT plans) were from institution 2. All VMAT plans, with the exceptions of 4 H&N VMAT plans were clinically acceptable showing global gamma passing rates with 2%/2 mm higher than 90% (Heilemann et al 2013). The global gamma passing rates with 2%/2 mm criteria of those plans were 88.2%, 81.6%, 79.3% and 71.5%, and were rejected for patient treatment. All VMAT plans in this study used two full arcs, 6 MV photon beams and Millennium 120™ MLC (Varian Medical Systems, Palo Alto, CA). The linacs of institution 1 and 2 were Trilogy® and Clinac iX® (Varian Medical Systems, Palo Alto, CA), respectively. Each VMAT plan was generated using the Eclipse™ system (Varian Medical Systems, Palo Alto, CA). Optimization was performed with the progressive resolution optimizer 3 algorithm (PRO3, ver.10, Varian Medical Systems, Palo Alto, CA). After optimization, dose distributions were calculated using the anisotropic analytic algorithm (AAA, ver.10, Varian Medical Systems, Palo Alto, CA) with a calculation grid of 2.5 mm. For H&N VMAT plans, a total of 3 target volumes (target67.5 Gy, target54 Gy and target48 Gy) or 2 target volumes (target67.5 Gy and target54 Gy) were treated with the simultaneous integrated boost (SIB) technique. The prescription doses of target67.5 Gy, target54 Gy and target48 Gy were 67.5 Gy (daily 2.25 Gy), 54 Gy (daily 1.8 Gy) and 48 Gy (daily 1.6 Gy), respectively. The prescription doses for prostate VMAT plans were 50.4 Gy (daily 1.8 Gy) with a primary plan and 30.6 Gy (daily 1.8 Gy) with a boost plan. Only primary plans were analyzed in this study. 2.3. Measure of VMAT delivery accuracy
A total of 3 types of verification methods for each VMAT plan were used to verify VMAT delivery accuracy in this study, which were the gamma-index method, analysis of mechanical parameter differences between plans versus linac log files and DynaLog files registered during delivery, and an analysis of differences in clinically relevant dose-volumetric parameters between plans versus the plans reconstructed with linac log files and DynaLog files. For the gamma-index method, both global and local gamma evaluations were performed with gamma criteria of 2%/2 mm, 2%/1 mm and 1%/2 mm using MapCHECK2™ detector array with MapPHAN™ (Sun Nuclear Corporation, Melbourne, FL). As recommended by recent studies on the gamma-index method for VMAT, gamma criteria of 3%/3 mm and 1%/1 mm were not used in this study (Poppe et al 2006, Heilemann et al 2013). To use 2%/1 mm gamma criterion, the verification plans for gamma evaluations were calculated with a calculation grid of 1 mm. To ensure accurate setup of the MapCHECK2™ detector array inserted into MapPHAN™, cone beam computed tomography (CBCT) images were acquired before measurements. The normalization dose for global gamma evaluation was the maximum 7109
J M Park et al
Phys. Med. Biol. 60 (2015) 7101
dose in this study. Values higher than 10% of the maximum dose were only evaluated with gamma-index method as is often cited in the literature (Gloi et al 2011, Heilemann et al 2013, Kim et al 2014). The gamma passing rates were calculated with SNC software (ver.6.1.2, Sun Nuclear Corporation, Melbourne, FL). The values of MLC positions, gantry angles and delivered MUs at each CP were acquired with dynamic log files and DynaLog files. Those log files were combined and reformatted as DICOM-RT files using an in-house program written in MATLAB (ver.8.1, Mathworks Inc., Natick, MA). The values of MLC positions, gantry angles and delivered MUs at each CP acquired with DICOM-RT formatted files were compared with those of original VMAT plans. The differences in mechanical parameters were calculated and averaged for each VMAT plan. The DICOM-RT files were imported in the Eclipse™ system and the dose-volumetric parameters were calculated with patient CT image sets under identical conditions as used for dose calculations of the original VMAT plans. The clinically relevant dose-volumetric parameters for target volumes were the dose received by 95% of target volume (D95%), D5%, minimum dose, maximum dose and mean dose. The dose-volumetric parameters for organs at risk (OARs) of prostate VMAT plans were D20% of rectal wall and bladder, D50% of femoral heads and the mean dose to rectal wall, bladder and femoral heads while those of H&N VMAT plans were the mean dose to each parotid gland and the maximum dose to spinal cord, brain stem, each lens, optic chiasm and each optic nerve. Those dose-volumetric parameters calculated with the plans reconstructed with log files were compared to those of original VMAT plans and the differences were calculated. A total of 35 kinds of dose-volumetric parameters including both prostate and H&N VMAT plans were evaluated in this study. 2.4. Correlation analysis
To evaluate the predictive power of MIc on VMAT delivery accuracy, Spearman’s rank correlation coefficients (rs) and corresponding p values were calculated between the values of MIc and (1) the gamma passing rates, (2) the differences in mechanical parameters between plans and log files and (3) the differences in dose-volumetric parameters between plans and the plans reconstructed with log files. Under the two-tailed unpaired parameter condition, p values for rs values were calculated. For comparison purpose, the values of MIt ( f = 0.5) were also calculated. With the values of MIt ( f = 0.5), the values of rs and corresponding p values to the gamma passing rates, mechanical parameter differences and dose-volumetric parameter differences were also calculated (Park et al 2014b). In addition, we calculated previously suggested modulation indices for VMAT, which were MCSv, LTMCS, MISPORT, PA, PI, PM, PMU and textural features of VMAT fluences such as contrast (d = 1), variance (d = 1) and contrast with edge-enhanced fluence (contrastedge (d = 1)) (Li and Xing 2013, Masi et al 2013, Du et al 2014, Park et al 2014d, 2015b). The values of rs and corresponding p values between each modulation index and the three types of verification methods for VMAT delivery accuracy were calculated. 3. Results 3.1. The values of modulation indices for VMAT plans
The values of various modulation indices including MIc are shown in table 2. All values of the modulation indices of prostate VMAT plans were different from those of H&N VMAT plans with statistical significances (always with p
