- Email: [email protected]
386 oral REAL-TIME VERIFICATION OF MLC-DRIVEN RADIOTHERAPY USING AN OPTICAL ATTENUATION-BASED FLUENCE MONITOR
W EDNESDAY, M AY 11, 2011
MU delivery error changed the TCP by 8.3% for the prostate plan. This plan passed a gamma criterion of 5%/4mm but failed at 4%/4mm. Effects on the oropharynx plan were comparable but slightly smaller as the TCP was higher so a lower gradient part of the dose response curve was being explored. A 4% MU delivery error in the oropharynx plan changed the TCP by 3%; this error was detected by a gamma criterion of 3%3 mm. Conclusions: The most significant biological effect was a reduction in TCP associated with MU delivery errors. For the plans tested, none of the errors had a significant effect on NTCP. A gamma threshold of 3%/3mm or less was required to ensure changes in TCP in the prostate were less than 6%. However, gamma criteria were insensitive to errors in mean dose and a criterion of 2% mean dose difference should be used in conjunction with the gamma criterion. With both criteria applied, a reduction in TCP of 4% would be identified in the prostate plan. 386 oral REAL-TIME VERIFICATION OF MLC-DRIVEN RADIOTHERAPY USING AN OPTICAL ATTENUATION–BASED FLUENCE MONITOR M. Goulet1 , L. Gingras1 , L. Beaulieu1 1 C ENTRE H OSPITALIER U NIVERSITAIRE DE Q UÉBEC, L’H ÔTEL -D IEU DE Q UÉBEC, Département de physique, Quebec, Canada Purpose: MLC-driven conformal radiotherapy modalities (e.g. IMRT, IMAT and SBRT) are more subject to delivery errors and dose calculation inaccuracies than standard modalities. Fluence monitoring during treatment delivery could reduce such errors by allowing an independent interface to quantify and assess measured difference between the delivered and planned treatment administration. We developed an optical attenuationbased detector to monitor fluence for the online quality control of radiotherapy delivery that can detected both errors in delivered fluence and individual leaf position. The purpose of the current study was to develop a theoretical framework of the invention and to evaluate experimentally the detector’s performance. Materials: We aligned sixty 27-cm scintillating fibers coupled to a photodetector via clear optical fibers in the direction of motion of each of the 60 leaf pairs of a 120 leaves Millenium MLC on a Varian Clinac iX, as shown on the included figure. The whole device is placed in the accessory tray of the linac so that fluence verification can take place during the treatment delivery. We developed a theoretical model to predict the intensity of light collected on each side of the scintillating fibers when placed under radiation fields of varying sizes, intensities, and positions. The model showed that both the central position of the radiation field on the fiber (xc) and the integral fluence passing through the fiber (int) could be assessed independently in a single measurement. We evaluated the performance of the prototype by 1) measuring the intrinsic variation of the measured values of xc and int as a function of field size, 2) measuring the impact on the measured values of xc and int of random leaf positioning errors introduced into IMRT fields and 3) measuring the influence of the fluence monitor on the incoming radiation beam.
Results: We observed a very low intrinsic dispersion, dominated by Poisson statistics, for both xc (standard deviations of less than 1 mm) and int (standard deviations of less than 0.25 %). When confronted with random leaf positioning errors from IMRT segments, int was highly sensitive to single errors as small as 1 mm at isocenter, while xc was sensitive to leaf pair translation errors of at least 2 mm at isocenter. The fluence monitor achieved a uniform beam transmission of 98.3%. Changes in percent depth-dose and relative dose profiles of square beams were found to be within 1%. Conclusions: Our study showed that an optical attenuationbased detector could be used to effectively monitor fluence during radiotherapy delivery. The performance of such a system would enable real-time quality control of the incident fluence in current MLC-driven treatments such as IMRT and future
P ROFFERED PAPER
S 153
adaptive radiotherapy procedures where new treatment plans will have to be delivered without passing thru the current standard quality control chain. 387 oral AN INDEPENDENT DOSE CALCULATION PROGRAM FOR THE CHECKING OF TOMOTHERAPY PLANS S. Thomas1 , K. Eyre1 , S. Tudor1 , J. Fairfoul1 1 A DDENBROOKE ’ S H OSPITAL, Medical Physics, Cambridge, United Kingdom Purpose: It is usual practice for all patients being treated on Tomotherapy to have individual delivery quality assurance (DQA) measurements to verify the dose delivered to a phantom. A major reason why it has not possible to relax this requirement has been the lack of an independent means of calculating the dose distribution. We have developed an independent calculation system which takes the patient-specific files from the TomoTherapy planning system and applies it to the patient CT data set in three dimensions. Materials: Beam data was measured to provide output factors, tissue phantom ratios, longitudinal beam profiles and in-plane beam profiles. They were measured on a TomoTherapy Hi-Art using A1SL chambers in a water tank. The HiArt delivers radiotherapy by rotating a beam around the patient whilst moving the couch longitudinally. The fluence is modulated with a set of binary collimators. The fluence pattern for each of 51 gantry positions, for each rotation, is saved in a sinogram file. The planning system also saves an XML file, that contains information such as the field width (defined in the direction of couch motion), the pitch, the speed of gantry rotation, the gantry start angle and the patient position at the start of irradiation. Each projection in the sinogram can be modelled as a set of segments. A program was written in Matlab to calculate the dose by interpolating into the measured beam data for each segment. The radiological path length for each calculation point is calculated by projecting through the CT images. Dose is calculated at 3375 calculation points, and compared with the doses in the DICOM dose cube exported by the Tomotherapy planning system. Results: The system was tested on a range of prostate and prostate+nodes plans. Calculation times range from 1.1 to 4.7 minutes, depending on the complexity of the sinogram. Averaged over all points within the 50% isodose surface, the mean dose difference between the two sets of calculated doses ranged from -1.3% to +0.6% for prostate plans, -0.3% to +2.4% for prosate+nodes plans. The Gamma index was used to compare dose distributions. A 4mm/4% Gamma index gave between 93% and 98% of points within the 50% isodose surface achieving Gamma<1 for prostate patients. Figure 1 shows a typical Gamma histogram. The Kappa and Box indices can be used as alternatives to Gamma. Simulating moves of 2% and 2mm causes the percentage passing to fall by 8-10%, demonstrating the use of the system in detecting simulated errors. Conclusions: Having a fast means of independently checking the dose distribution provides a useful additional safeguard against errors. The use of beam data derived from measurement ensures independence from the planning system. The system achieves a level of accuracy sufficient to serve as a means of detecting errors in the plan. We are currently undergoing final testing and evaluation of the system, prior to a decision of whether to reduce the frequency of DQA measurements.
MU delivery error changed the TCP by 8.3% for the prostate plan. This plan passed a gamma criterion of 5%/4mm but failed at 4%/4mm. Effects on the oropharynx plan were comparable but slightly smaller as the TCP was higher so a lower gradient part of the dose response curve was being explored. A 4% MU delivery error in the oropharynx plan changed the TCP by 3%; this error was detected by a gamma criterion of 3%3 mm. Conclusions: The most significant biological effect was a reduction in TCP associated with MU delivery errors. For the plans tested, none of the errors had a significant effect on NTCP. A gamma threshold of 3%/3mm or less was required to ensure changes in TCP in the prostate were less than 6%. However, gamma criteria were insensitive to errors in mean dose and a criterion of 2% mean dose difference should be used in conjunction with the gamma criterion. With both criteria applied, a reduction in TCP of 4% would be identified in the prostate plan. 386 oral REAL-TIME VERIFICATION OF MLC-DRIVEN RADIOTHERAPY USING AN OPTICAL ATTENUATION–BASED FLUENCE MONITOR M. Goulet1 , L. Gingras1 , L. Beaulieu1 1 C ENTRE H OSPITALIER U NIVERSITAIRE DE Q UÉBEC, L’H ÔTEL -D IEU DE Q UÉBEC, Département de physique, Quebec, Canada Purpose: MLC-driven conformal radiotherapy modalities (e.g. IMRT, IMAT and SBRT) are more subject to delivery errors and dose calculation inaccuracies than standard modalities. Fluence monitoring during treatment delivery could reduce such errors by allowing an independent interface to quantify and assess measured difference between the delivered and planned treatment administration. We developed an optical attenuationbased detector to monitor fluence for the online quality control of radiotherapy delivery that can detected both errors in delivered fluence and individual leaf position. The purpose of the current study was to develop a theoretical framework of the invention and to evaluate experimentally the detector’s performance. Materials: We aligned sixty 27-cm scintillating fibers coupled to a photodetector via clear optical fibers in the direction of motion of each of the 60 leaf pairs of a 120 leaves Millenium MLC on a Varian Clinac iX, as shown on the included figure. The whole device is placed in the accessory tray of the linac so that fluence verification can take place during the treatment delivery. We developed a theoretical model to predict the intensity of light collected on each side of the scintillating fibers when placed under radiation fields of varying sizes, intensities, and positions. The model showed that both the central position of the radiation field on the fiber (xc) and the integral fluence passing through the fiber (int) could be assessed independently in a single measurement. We evaluated the performance of the prototype by 1) measuring the intrinsic variation of the measured values of xc and int as a function of field size, 2) measuring the impact on the measured values of xc and int of random leaf positioning errors introduced into IMRT fields and 3) measuring the influence of the fluence monitor on the incoming radiation beam.
Results: We observed a very low intrinsic dispersion, dominated by Poisson statistics, for both xc (standard deviations of less than 1 mm) and int (standard deviations of less than 0.25 %). When confronted with random leaf positioning errors from IMRT segments, int was highly sensitive to single errors as small as 1 mm at isocenter, while xc was sensitive to leaf pair translation errors of at least 2 mm at isocenter. The fluence monitor achieved a uniform beam transmission of 98.3%. Changes in percent depth-dose and relative dose profiles of square beams were found to be within 1%. Conclusions: Our study showed that an optical attenuationbased detector could be used to effectively monitor fluence during radiotherapy delivery. The performance of such a system would enable real-time quality control of the incident fluence in current MLC-driven treatments such as IMRT and future
P ROFFERED PAPER
S 153
adaptive radiotherapy procedures where new treatment plans will have to be delivered without passing thru the current standard quality control chain. 387 oral AN INDEPENDENT DOSE CALCULATION PROGRAM FOR THE CHECKING OF TOMOTHERAPY PLANS S. Thomas1 , K. Eyre1 , S. Tudor1 , J. Fairfoul1 1 A DDENBROOKE ’ S H OSPITAL, Medical Physics, Cambridge, United Kingdom Purpose: It is usual practice for all patients being treated on Tomotherapy to have individual delivery quality assurance (DQA) measurements to verify the dose delivered to a phantom. A major reason why it has not possible to relax this requirement has been the lack of an independent means of calculating the dose distribution. We have developed an independent calculation system which takes the patient-specific files from the TomoTherapy planning system and applies it to the patient CT data set in three dimensions. Materials: Beam data was measured to provide output factors, tissue phantom ratios, longitudinal beam profiles and in-plane beam profiles. They were measured on a TomoTherapy Hi-Art using A1SL chambers in a water tank. The HiArt delivers radiotherapy by rotating a beam around the patient whilst moving the couch longitudinally. The fluence is modulated with a set of binary collimators. The fluence pattern for each of 51 gantry positions, for each rotation, is saved in a sinogram file. The planning system also saves an XML file, that contains information such as the field width (defined in the direction of couch motion), the pitch, the speed of gantry rotation, the gantry start angle and the patient position at the start of irradiation. Each projection in the sinogram can be modelled as a set of segments. A program was written in Matlab to calculate the dose by interpolating into the measured beam data for each segment. The radiological path length for each calculation point is calculated by projecting through the CT images. Dose is calculated at 3375 calculation points, and compared with the doses in the DICOM dose cube exported by the Tomotherapy planning system. Results: The system was tested on a range of prostate and prostate+nodes plans. Calculation times range from 1.1 to 4.7 minutes, depending on the complexity of the sinogram. Averaged over all points within the 50% isodose surface, the mean dose difference between the two sets of calculated doses ranged from -1.3% to +0.6% for prostate plans, -0.3% to +2.4% for prosate+nodes plans. The Gamma index was used to compare dose distributions. A 4mm/4% Gamma index gave between 93% and 98% of points within the 50% isodose surface achieving Gamma<1 for prostate patients. Figure 1 shows a typical Gamma histogram. The Kappa and Box indices can be used as alternatives to Gamma. Simulating moves of 2% and 2mm causes the percentage passing to fall by 8-10%, demonstrating the use of the system in detecting simulated errors. Conclusions: Having a fast means of independently checking the dose distribution provides a useful additional safeguard against errors. The use of beam data derived from measurement ensures independence from the planning system. The system achieves a level of accuracy sufficient to serve as a means of detecting errors in the plan. We are currently undergoing final testing and evaluation of the system, prior to a decision of whether to reduce the frequency of DQA measurements.