Page 47 - FiM_03_2016

Measurements were performed by using TrueBeam STx accelerator equipped with EPID aSi1000 (Varian, Palo Alto, CA) and PerFraction (PF) software (Sun Nuclear Corporation, Melbourne, FL). PF is a commercial EPID-based dosimetry software, which allows performing transit dosimetry, to provide an independent daily verification of the treatment. Performance of the EPID detector and of the PF software on anthropomorphic phantom was studied, simulating 17 perturbations of the reference VMAT plan. Systematic variations in dose values (1%-5% output variation), shifts (2,5- 11 mm in anterior direction), anatomical variations (adding bolus over phantom), and MLC positioning (locked leaf position for different arc extensions) were applied. The difference in local and global gamma pass rate (%GP) between the no-error and error-simulated measurements with 1%/1mm, 2%/2 mm and 3%/3 mm tolerances was calculated. The clinical impact of these errors was also analyzed through the calculation of the difference between the reference DVH and the perturbed DVH (%DE). We defined as clinically meaningful a variation higher than 3% between calculated and perturbed doses. A value of %GP equal to 95% and 90% and %DE equal to 3% were used as thresholds to calculate sensitivity and specificity.
Nineteen patients undergoing VMAT treatment were enrolled in this study. We performed first a pre-treatment QA verification: patient plan was delivered in QA mode and acquired by the EPID in continuous mode, SID = 150cm. During the delivery, the couch was removed. Once performed the pre-treatment QA verification the patient started the treatment: before each fraction a cone-beam CT was delivered to reach the best setup; during the delivery of each fraction the EPID acquired the transmitted dose in cinematic modality. The PerFraction algorithm takes 10 to 30 minutes to recalculate the dose on the original patient CT. No information on patient setup, such as the cone- beam CT, are taken into account in this dose calculation. Recalculated dose distribution was exported to another software (3DVH, Sun Nuclear Corporation, Melbourne, FL) where it was compared with the dose distribution calculated by the TPS, used as reference. Dose at target was compared with the dosimetric parameter D2%, D95% and Dmean. We evaluated the dose of two PTVs if the plan presented more than one target and of one PTV and one CTV otherwise. For each plan, the two more critical OARs were selected and up to three dosimetric parameters were chosen, for each OAR, in accordance with QUANTEC [3]. A total of 12 parameters were evaluated for every plan. The parameter percentage relative variation (%DE) was calculated between the reference plan and the recalculated one. We defined as clinically meaningful a variation higher than 3% between DVH calculated and recalculated doses. Finally, we compared the recalculated dose parameters with the QUANTEC ones. Moreover, starting from the second fraction, the recalculated dose was compared, as described before, with the first fraction recalculated dose. By this analysis it was evaluated the constancy of the delivery through several fractions.
Tesi di Specialità
39