IMRT QA Analysis: What to Do with Unresolved Discrepancies?
R.A Price,J Li,J Yang,S.W McNeeley,L Chen,E Fourkal,M Ding,W Xiong,L Qin,G Mora,C.C Ma
DOI: https://doi.org/10.1016/s0360-3016(03)01099-x
2003-01-01
Abstract:Purpose/Objective: To date, we have treated approximately 800 patients with IMRT at this institution using the Siemens Primus or Primart linear accelerators and the SMLC delivery technique. Prior to the initial patient treatment, each plan is evaluated and a series of QA tests are performed including absolute dose verification. This verification is a comparison between the dose calculated in phantom by the treatment planning system and the dose measured in phantom during irradiation. Our clinical acceptance criteria dictate that the values should agree to within ± 3% for treatment to begin. Values exceeding this limit trigger additional analysis, measurement and double-checking prior to patient treatment. An example is given where the mean deviation value was −6.5% with a maximum deviation measured of −8.8%. The focus of this work is to analyze the accuracy of our dosimetry QA procedure and address discrepancies that are significantly greater than ± 3%.Materials/Methods: All IMRT treatment plans are generated using the Corvus treatment planning system. Once a satisfactory plan is obtained a hybrid plan is generated. This plan utilizes the leaf sequences and associated MU designed for the patient plan and calculates dose on a virtual phantom. The physical phantom is then irradiated under the same conditions using the same leaf sequences. An absolute dose comparison is made by comparing dose in the physical phantom as measured using a 0.125cc PTW ionization chamber with dose predicted at the same point(s) in the virtual phantom by the treatment planning system. Agreement to within ± 3% indicates that the plan is appropriately deliverable as far as absolute dose is concerned. Values exceeding this limit prompt additional investigation. This investigation includes reassessing the treatment plan and hybrid plan for errors in data extraction, investigation of input data into the R&V system for accuracy and completeness, and investigation of deliverability with respect to beam placement. The phantom irradiation is then repeated to eliminate errors due to improper setup. Discrepancies outside a ± 4% window are verified using Monte Carlo simulations of both the dose in the virtual phantom and dose in the original patient CT data set. With agreement between the original Corvus calculated relative dose distributions and the distributions generated in film and Monte Carlo simulations, the MU can be scaled appropriately to bring the absolute dose into our acceptable range. When agreement is not present the individual patient plans are regenerated with changes in input parameters and/or beam directions.Results: Upon evaluation of our absolute dose data it was found that the agreement between measured dose and dose predicted by the planning system in phantom was within ± 1% for 37.9% of the cases, ± 2% for 68.6% and ± 3% for 93.6% of the cases. The discrepancies were more than 3% and more than 4% in only 6.4% and less than 1% of all cases, respectively. Monte Carlo calculation of absolute dose compares to within 2% of dose in phantom or the patient CT data set in the non-gradient portion of the dose distribution associated with the target as tested for 20 prostate cases.Conclusions: There appear to be some intensity maps that are not deliverable to within our clinical acceptance criteria even though the planning system generated a leaf sequence and associated MU set. Verification of absolute dose as well as the relative dose distribution is essential especially in these cases. While MU scaling may result in acceptable delivery, verification of the resultant dose distribution is needed since the Siemens linacs allow for integer MU delivery only. Monte Carlo can serve as an independent dose verification tool in the QA process. Evaluation of different leaf sequencing options is made for these cases and comparisons made using both measurement and Monte Carlo simulations. Purpose/Objective: To date, we have treated approximately 800 patients with IMRT at this institution using the Siemens Primus or Primart linear accelerators and the SMLC delivery technique. Prior to the initial patient treatment, each plan is evaluated and a series of QA tests are performed including absolute dose verification. This verification is a comparison between the dose calculated in phantom by the treatment planning system and the dose measured in phantom during irradiation. Our clinical acceptance criteria dictate that the values should agree to within ± 3% for treatment to begin. Values exceeding this limit trigger additional analysis, measurement and double-checking prior to patient treatment. An example is given where the mean deviation value was −6.5% with a maximum deviation measured of −8.8%. The focus of this work is to analyze the accuracy of our dosimetry QA procedure and address discrepancies that are significantly greater than ± 3%. Materials/Methods: All IMRT treatment plans are generated using the Corvus treatment planning system. Once a satisfactory plan is obtained a hybrid plan is generated. This plan utilizes the leaf sequences and associated MU designed for the patient plan and calculates dose on a virtual phantom. The physical phantom is then irradiated under the same conditions using the same leaf sequences. An absolute dose comparison is made by comparing dose in the physical phantom as measured using a 0.125cc PTW ionization chamber with dose predicted at the same point(s) in the virtual phantom by the treatment planning system. Agreement to within ± 3% indicates that the plan is appropriately deliverable as far as absolute dose is concerned. Values exceeding this limit prompt additional investigation. This investigation includes reassessing the treatment plan and hybrid plan for errors in data extraction, investigation of input data into the R&V system for accuracy and completeness, and investigation of deliverability with respect to beam placement. The phantom irradiation is then repeated to eliminate errors due to improper setup. Discrepancies outside a ± 4% window are verified using Monte Carlo simulations of both the dose in the virtual phantom and dose in the original patient CT data set. With agreement between the original Corvus calculated relative dose distributions and the distributions generated in film and Monte Carlo simulations, the MU can be scaled appropriately to bring the absolute dose into our acceptable range. When agreement is not present the individual patient plans are regenerated with changes in input parameters and/or beam directions. Results: Upon evaluation of our absolute dose data it was found that the agreement between measured dose and dose predicted by the planning system in phantom was within ± 1% for 37.9% of the cases, ± 2% for 68.6% and ± 3% for 93.6% of the cases. The discrepancies were more than 3% and more than 4% in only 6.4% and less than 1% of all cases, respectively. Monte Carlo calculation of absolute dose compares to within 2% of dose in phantom or the patient CT data set in the non-gradient portion of the dose distribution associated with the target as tested for 20 prostate cases. Conclusions: There appear to be some intensity maps that are not deliverable to within our clinical acceptance criteria even though the planning system generated a leaf sequence and associated MU set. Verification of absolute dose as well as the relative dose distribution is essential especially in these cases. While MU scaling may result in acceptable delivery, verification of the resultant dose distribution is needed since the Siemens linacs allow for integer MU delivery only. Monte Carlo can serve as an independent dose verification tool in the QA process. Evaluation of different leaf sequencing options is made for these cases and comparisons made using both measurement and Monte Carlo simulations.