- Email: [email protected]
OC-0229: EPID dose response in the MR-Linac with and without presence of a magnetic field
S114 ESTRO 36 _______________________________________________________________________________________________ OC-0228 DVH criteria for prostate in vivo EPID dosimetry R.F.M. Van Oers1, E. Van der Bijl1, I. Olaciregui-Ruiz1, A. Mans1 1 Netherlands Cancer Institute, Radiation Oncology, Amsterdam, The Netherlands Purpose or Objective In our department in vivo EPID dosimetry is used for dose verification of all treatment plans. The algorithm uses EPID images acquired behind the patient to reconstruct the in vivo 3D dose distribution. This is then automatically compared to the planned dose distribution and alerts are generated when deviations are detected. These alerts are based on γ-analysis. Gamma values combine dose-difference and distance-to-agreement in a single metric, but this metric contains no information on the clinical relevance of deviations. Furthermore, γanalysis can be insensitive to systematic under- or overdoses in plans with inhomogeneous dose distributions. Dose-volume histograms (DVHs) are widely used for evaluation of treatment plans, and readily understood by clinicians. Moreover, differences in DVH parameters can be linked more straightforwardly to clinical relevance. This makes DVH-based criteria an attractive alternative to γ-criteria for in vivo EPID dosimetry alerts. In this study we investigated the correlation between γand DVH-parameters of the PTV for prostate treatments, and compared the alert rates for different criteria in order to propose a suitable set of DVH-criteria for clinical implementation. Material and Methods In vivo 3D dose distributions were reconstructed for the first three fractions of 95 prostate VMAT treatments, and then averaged for the evaluation of each treatment. The γ-analysis was done with global 3%/3mm settings. DVHs were obtained for the PTV (prostate + seminal vesicles). Calculated γ-parameters were mean γ, the near-maximum γ (γ1%), and the γ-passrate (γ%<1); our current criteria also include the isocenter dose difference (ΔDisoc). Calculated DVH-parameters were the difference in nearmaximum dose (ΔD2), median dose (ΔD50), and nearminimum dose (ΔD98). We obtained alert rates for different sets of criteria on these DVH-parameters. These were compared to alert rates for the current γ-based criteria. There are two alert levels, higher-priority “error” and lower-priority “warning”. Results The strongest correlation was found between γ-mean and |ΔD50|, Pearson’s r=0.95. All other γ- and DVH parameters were also strongly correlated, with r values around 0.85.
Figure 1: Relation between γ-mean and ΔD50 for the analyzed prostate VMAT plans. The indicated alerts are generated by the “error level” sets of γ- and DVH-criteria. Table 1 shows alert rates for different sets of γ- and DVHcriteria. The two highlighted sets of γ-criteria are the ones currently used in our clinic to generate alerts for prostate
treatments, juxtaposed with sets of DVH-criteria of similar alert rate.
Table 1: Alert rates (% of treatments) for different sets of γ- and DVH-criteria. The top set of γ-criteria corresponds to “warning level” alerts, the bottom set corresponds to “error level” alerts. Conclusion A strong correlation was found between γ- and DVHparameters of the PTV; a set of DVH-criteria that performs comparably to the current γ-criteria can easily be chosen. OC-0229 EPID dose response in the MR-Linac with and without presence of a magnetic field I. Torres Xirau1, I. Olaciregui-Ruiz1, B. J. Mijnheer1, U. A. van der Heide1, A. Mans1 1 Netherlands Cancer Institute Antoni van Leeuwenhoek Hospital, Department of Radiation Oncology, Amsterdam, The Netherlands Purpose or Objective Image-guided radiotherapy systems are being investigated and clinically implemented aiming for online and real-time adaptation of the treatment plan. The use of Electronic Portal Imaging Devices (EPIDs) for independent in vivo dose verification in the Elekta MR-Linac is being developed. One of the challenges for MR-Linac portal dosimetry is the presence of a small magnetic field at the EPID level. In the presence of a magnetic field, the secondary electrons that actually deposit the dose in the scintillator of the EPID will be affected by the Lorentz force possibly leading to a B-field induced dose redistribution. The aim of this study was to analyze and quantify the effects of the B-field on the EPID images acquired on the Elekta MR-Linac. Material and Methods The Elekta/Philips MR-Linac combines a 1.5T magnetic resonance imaging scanner with a linear accelerator and is equipped with an on-board EPID. A magnetometer (MetroLab THM1176) was used to measure the strength of the magnetic B-field at the surface of the EPID. To assess the reproducibility of the panel readouts, a 10x10 cm2 field was irradiated 10 times in two consecutive days and the value of the on-axis region (averaged 5x5 pixels) of EPID images was recorded. During the installation of the MR-Linac in our institute, EPID images were acquired before the B-field was ramped up and repeated with Bfield one month later. To study the on-axis response of the EPID as function of field size with and without the magnetic B-field, square fields were irradiated with field sizes varying from 2 to 20 cm. Furthermore, EPID images acquired with and without B-field were compared by means of a 2-D γ-analysis (local 2%,1mm, 20% isodose) and X-Y EPID lateral profiles were compared by visual inspection. Results The magnetic field measured on top of the panel did not exceed 2.5 mT, yielding an electron trajectory radius of approximately 1.20 m. The reproducibility of EPID central axis values for ten irradiated 10x10 cm2 fields was 0.3% (1 SD). The normalized on-axis EPID response as function of field size with and without the presence of the magnetic field is shown in Figure 1 together with their ratio.
S115 ESTRO 36 _______________________________________________________________________________________________
Figure 2 shows the comparison between the normalized lateral X and Y profiles of EPID images acquired with and without the B-field (3x3,5x5,8x8,10x10,12x12,15x15,20x20 cm2). More than 99% of the points showed local deviations smaller than 2% for the X and Y profiles.
combined to treat large areas) necessary for large H&N and pelvic treatments. Material and Methods As of November 2015, 43 patients were treated on G2 for a total of 74 plans (21 SFUD, 51 IMPT and 3 SIB) and 248 fields (average of 4 fields per plan), of which 26 fields were patched. Parameters recorded during the treatment delivery of these patients (spot positions, MU’s per pencil beam, couch and gantry position) are stored into a log file and used to reconstruct the 3D dose distribution by an inhouse developed Independent Dose Calculation software (Meier et al. 2015). A MATLAB script calculates the dose metrics by comparing the reconstructed to the nominal dose distribution. These metrics include the maximum, minimum and mean dose differences as well as the percentage of voxels within +/- 1% of the nominal dose (pass rate). Results Table 1 shows the results of the log file analysis. Interestingly, and despite the typically higher modulation for IMPT, the average pass rate for both SFUD and IMPT is similar, with the 95% percentile actually being a little better for IMPT. In addition, complex plans with steep infield dose gradients, such as SIB treatments, also had pass rates >99%. Nevertheless, highly modulated plans can have larger local dose differences as seen by the larger max dose deviation in Table 1 and demonstrated for a specific case in Figure 1. Hence, attention should be paid to the location of isolated, highly weighted spots.
The 2-D γ-analysis showed that the averaged γmean was 0.42 ± 0.16 and the %γ≤1 was 96.6 ± 4.5. Conclusion EPID images acquired with and without B-field are virtually identical, indicating that the presence of a small (2.5 mT) magnetic field at the EPID level in the MR-Linac should not become an impediment for the implementation of EPID dosimetry in the MR-Linac. Acknowledgements This research was partly sponsored by Elekta AB, Stockholm, Sweden. The authors would like to thank Robert Spaninks (Elekta) for assistance with the measurements. OC-0230 Treatment log files as a tool to identify inaccuracies in scanned proton beam delivery and planning M. Belosi1, R. Van der Meer1, P. Garcia de Acilu Laa2, A. Bolsi1, D. Weber1, A. Lomax1 1 Paul Scherrer Institute, Centre for Proton Therapy, Villigen PSI, Switzerland 2 Hospital Universitario Puerta del Sur Hospitales de Madrid, Radiofisica Hospitalitaria, Madrid, Spain Purpose or Objective Dose distributions delivered at Gantry2 (G2) at the Paul Scherrer Institut (PSI) can be reconstructed on the patient anatomy based on machine log files. These dose reconstructions are a powerful tool in identifying potential issues related to the integrity of the patients’ dose delivery, as has already been demonstrated for a first series of patients treated in G2 for skull base chordomas (Scandurra et al. 2016). Here, such calculations have been extended by investigating their dependency on planning technique (e.g. SFUD vs IMPT, field direction etc) and on couch position. The latter is crucial for quality assurance of the delivery of patched fields (different sub-fields
Finally, the results of the first patched field treatments (2 to 4 patches per field) did not show any evidence of dose deviations at the interface between patches. Conclusion 3D dose reconstruction using treatment log files is a powerful tool to identify delivery problems and trends, and to improve planning robustness. Further effort should be invested in order to predict field robustness to delivery fluctuations before the clinical delivery of the plan as part of the plan’s specific QA. OC-0231 The suitability of radiochromic film in 0.35T magnetic field CO-60 compared with conventional 6MV D.L.J. Barten1, L.J. Van Battum1, D. Hoffmans1, S. Heukelom1 1 VUMC, Radiotherapie, Amsterdam, The Netherlands
Figure 1: Relation between γ-mean and ΔD50 for the analyzed prostate VMAT plans. The indicated alerts are generated by the “error level” sets of γ- and DVH-criteria. Table 1 shows alert rates for different sets of γ- and DVHcriteria. The two highlighted sets of γ-criteria are the ones currently used in our clinic to generate alerts for prostate
treatments, juxtaposed with sets of DVH-criteria of similar alert rate.
Table 1: Alert rates (% of treatments) for different sets of γ- and DVH-criteria. The top set of γ-criteria corresponds to “warning level” alerts, the bottom set corresponds to “error level” alerts. Conclusion A strong correlation was found between γ- and DVHparameters of the PTV; a set of DVH-criteria that performs comparably to the current γ-criteria can easily be chosen. OC-0229 EPID dose response in the MR-Linac with and without presence of a magnetic field I. Torres Xirau1, I. Olaciregui-Ruiz1, B. J. Mijnheer1, U. A. van der Heide1, A. Mans1 1 Netherlands Cancer Institute Antoni van Leeuwenhoek Hospital, Department of Radiation Oncology, Amsterdam, The Netherlands Purpose or Objective Image-guided radiotherapy systems are being investigated and clinically implemented aiming for online and real-time adaptation of the treatment plan. The use of Electronic Portal Imaging Devices (EPIDs) for independent in vivo dose verification in the Elekta MR-Linac is being developed. One of the challenges for MR-Linac portal dosimetry is the presence of a small magnetic field at the EPID level. In the presence of a magnetic field, the secondary electrons that actually deposit the dose in the scintillator of the EPID will be affected by the Lorentz force possibly leading to a B-field induced dose redistribution. The aim of this study was to analyze and quantify the effects of the B-field on the EPID images acquired on the Elekta MR-Linac. Material and Methods The Elekta/Philips MR-Linac combines a 1.5T magnetic resonance imaging scanner with a linear accelerator and is equipped with an on-board EPID. A magnetometer (MetroLab THM1176) was used to measure the strength of the magnetic B-field at the surface of the EPID. To assess the reproducibility of the panel readouts, a 10x10 cm2 field was irradiated 10 times in two consecutive days and the value of the on-axis region (averaged 5x5 pixels) of EPID images was recorded. During the installation of the MR-Linac in our institute, EPID images were acquired before the B-field was ramped up and repeated with Bfield one month later. To study the on-axis response of the EPID as function of field size with and without the magnetic B-field, square fields were irradiated with field sizes varying from 2 to 20 cm. Furthermore, EPID images acquired with and without B-field were compared by means of a 2-D γ-analysis (local 2%,1mm, 20% isodose) and X-Y EPID lateral profiles were compared by visual inspection. Results The magnetic field measured on top of the panel did not exceed 2.5 mT, yielding an electron trajectory radius of approximately 1.20 m. The reproducibility of EPID central axis values for ten irradiated 10x10 cm2 fields was 0.3% (1 SD). The normalized on-axis EPID response as function of field size with and without the presence of the magnetic field is shown in Figure 1 together with their ratio.
S115 ESTRO 36 _______________________________________________________________________________________________
Figure 2 shows the comparison between the normalized lateral X and Y profiles of EPID images acquired with and without the B-field (3x3,5x5,8x8,10x10,12x12,15x15,20x20 cm2). More than 99% of the points showed local deviations smaller than 2% for the X and Y profiles.
combined to treat large areas) necessary for large H&N and pelvic treatments. Material and Methods As of November 2015, 43 patients were treated on G2 for a total of 74 plans (21 SFUD, 51 IMPT and 3 SIB) and 248 fields (average of 4 fields per plan), of which 26 fields were patched. Parameters recorded during the treatment delivery of these patients (spot positions, MU’s per pencil beam, couch and gantry position) are stored into a log file and used to reconstruct the 3D dose distribution by an inhouse developed Independent Dose Calculation software (Meier et al. 2015). A MATLAB script calculates the dose metrics by comparing the reconstructed to the nominal dose distribution. These metrics include the maximum, minimum and mean dose differences as well as the percentage of voxels within +/- 1% of the nominal dose (pass rate). Results Table 1 shows the results of the log file analysis. Interestingly, and despite the typically higher modulation for IMPT, the average pass rate for both SFUD and IMPT is similar, with the 95% percentile actually being a little better for IMPT. In addition, complex plans with steep infield dose gradients, such as SIB treatments, also had pass rates >99%. Nevertheless, highly modulated plans can have larger local dose differences as seen by the larger max dose deviation in Table 1 and demonstrated for a specific case in Figure 1. Hence, attention should be paid to the location of isolated, highly weighted spots.
The 2-D γ-analysis showed that the averaged γmean was 0.42 ± 0.16 and the %γ≤1 was 96.6 ± 4.5. Conclusion EPID images acquired with and without B-field are virtually identical, indicating that the presence of a small (2.5 mT) magnetic field at the EPID level in the MR-Linac should not become an impediment for the implementation of EPID dosimetry in the MR-Linac. Acknowledgements This research was partly sponsored by Elekta AB, Stockholm, Sweden. The authors would like to thank Robert Spaninks (Elekta) for assistance with the measurements. OC-0230 Treatment log files as a tool to identify inaccuracies in scanned proton beam delivery and planning M. Belosi1, R. Van der Meer1, P. Garcia de Acilu Laa2, A. Bolsi1, D. Weber1, A. Lomax1 1 Paul Scherrer Institute, Centre for Proton Therapy, Villigen PSI, Switzerland 2 Hospital Universitario Puerta del Sur Hospitales de Madrid, Radiofisica Hospitalitaria, Madrid, Spain Purpose or Objective Dose distributions delivered at Gantry2 (G2) at the Paul Scherrer Institut (PSI) can be reconstructed on the patient anatomy based on machine log files. These dose reconstructions are a powerful tool in identifying potential issues related to the integrity of the patients’ dose delivery, as has already been demonstrated for a first series of patients treated in G2 for skull base chordomas (Scandurra et al. 2016). Here, such calculations have been extended by investigating their dependency on planning technique (e.g. SFUD vs IMPT, field direction etc) and on couch position. The latter is crucial for quality assurance of the delivery of patched fields (different sub-fields
Finally, the results of the first patched field treatments (2 to 4 patches per field) did not show any evidence of dose deviations at the interface between patches. Conclusion 3D dose reconstruction using treatment log files is a powerful tool to identify delivery problems and trends, and to improve planning robustness. Further effort should be invested in order to predict field robustness to delivery fluctuations before the clinical delivery of the plan as part of the plan’s specific QA. OC-0231 The suitability of radiochromic film in 0.35T magnetic field CO-60 compared with conventional 6MV D.L.J. Barten1, L.J. Van Battum1, D. Hoffmans1, S. Heukelom1 1 VUMC, Radiotherapie, Amsterdam, The Netherlands