ESTRO 2020 Abstract book

S1042 ESTRO 2020

introduction in our department, and along a period of 7 months. Material and Methods The duration of a treatment session is defined in this work as the time from the moment the patient enters the treatment room, to the moment the patient leaves the treatment room (that is, the door-to-door time). This time includes patient setup, image acquisition, online couch correction, beam-on time, and the time needed to leave the treatment room. We studied the duration of treatment sessions as a function of the treatment site, the modality (VMAT or IMRT), and the time from the first day of treatments with the Halcyon 2.0 system in our department. The analysis considered 4569 treatment sessions, delivered from Mar 6, 2019 to Oct 2, 2019, and corresponding to 223 patients. All sessions included an image acquisition (kVCBCT or an orthogonal MV/MV pair) and an online image-guided correction. Data of treatment times was extracted from the Aria 15.6 database (Varian Medical Systems) using the Aura reporting tool. Treatment duration was derived from Aria data as the difference between the closing time of the plan at the linac console, and the corresponding opening plan time. The check the adequacy of this metric to evaluate the door-to-door time previously defined, we measured manually the actual door-to-door time for a sample of 96 treatment sessions. Site data was taken from an in-house database of pre- treatment plan QA. The data analysis was done with R The treatment duration as extracted from Aria needed to be corrected by a factor of 0.85±0.10 to obtain the actual door-to-door time. Boxplots in figure 1 show the corrected door-to-door times of the analyzed sessions, grouped by sites and modality. The number of analyzed sessions by site ranged from 65 for bladder to 1258 for prostate. The mean values of door- to-door time for all sites ranged from 6.2 min for CNS to 8.7 min for breast. The 90 th percentile values of door-to- door time for all sites ranged from 8.6 min for CNS to 11.8 min for breast. scripts. Results

To validate the freely available Pylinac software platform (https://pylinac.readthedocs.io/en/latest/) to be used for Dynalog file analysis. Material and Methods The AAPM Task Group 142 report [Med Phys. 2009 Sep; 36(9): 4197-212] recommends a MLC log file evaluation for moving-window IMRT to check the MLC positional accuracy. This document establishes a tolerance of 0.35 cm for the maximum and 95 th percentile of the positional errors. A Varian Clinac 2100 CD equipped with the Millennium 120 MLC was used (Varian Medical Systems, Palo Alto, CA). The MLC controller generates a pair of log files (Dynalog files, one per MLC bank) for each dynamic MLC delivery. The Dynalogs files can be analyzed using the DynaLog File Viewer (DFV, version 7.0.1.17) tool provided by Varian. This software creates an error histogram showing all the leaf position deviations, error RMS showing the calculated root mean square error for leaf deviations, and beam hold- off and beam-on plots. On the other hand, the Pylinac platform provides the Log Analyzer module for analysis of the Dynalog files. Dynalog files corresponding to 107 sliding-window fields of clinical treatments were randomly selected from our database. They were analyzed using the Pylinac version 2.2.7 and the DFV softwares, taking the DFV as the gold standard. The reported values for both softwares were compared (paired two-tailed Student's t- test) for the following metrics: 1) maximum RMS error (Max_RMS); 2) average RMS error per MLC bank (AvgRMS_A, AvgRMS_B); 3) 95 th percentile leaf error (95 th ), and 4) number of beam hold-offs. Agreement between both softwares was also assessed using the Bland-Altman method. While Pylinac provides the leaf error 95 th percentile value, DFV reports the bin containing the 95 th percentile from the leaf error histogram constructed with a 0.05 cm binning and a total range from 0 to 1 cm. Therefore, we checked whether the Pylinac 95 th percentile value lied or not in the 95 th bin reported by DFV. Results No statistically significant differences (p > 0.05) were found in the values of Max_RMS, AvgRMS_A and AvgRMS_B reported by Pylinac and DFV (mean ± SD): (0.00 ± 0.01) mm, (0.00 ± 0.11) mm and (0.02 ± 0.17) mm, respectively. According to the Bland-Altman analysis, the limits of agreement between both softwares were: [-0.02 mm, 0.02 mm], [-0.21 mm, -0.22 mm] and [-0.36 mm, 0.33 mm] for Max_RMS, AvgRMS_A and AvgRMS_B, respectively. Regarding to the 95 th percentile metric, Pylinac values were in congruence with the 95 th bins reported by DFV in all cases. Same number of beam hold-offs was found with both softwares in all cases. Conclusion Pylinac is fine to analyze the metrics required by the AAPM Task Group 142 report for Dynalog QA (maximum RMS error and the leaf error 95 th percentile) and may replace the DFV software. PO-1778 Analysis of treatment times with a Halcyon 2.0 linac M. Hermida Lopez 1 , D. García Relancio 2 , M. Comino Muriano 2 , B. Pérez Esteve 2 , E. Castillo Elías 2 , Y. Carrera García 2 , J. Giralt 2 1 Hospital Universitario Vall d'Hebron, Servei de Fisica i Protecció Radiològica, Barcelona, Spain ; 2 Hospital Universitario Vall d'Hebron, Servei d'Oncologia Radioteràpica, Barcelona, Spain Purpose or Objective The Halcyon linac was recently introduced in the market by Varian Medical Systems. According to the manufacturer, the system is designed to improve the clinical workflow, allowing efficient online image-guided radiotherapy with IMRT and VMAT techniques. This works measures and analyzes the duration of treatment sessions on a Halcyon 2.0 unit since its

Figure 2 shows the mean and standard deviation of the door-to-door times for each day of treatments in the analyzed period. After the first 11 days of treatments with a progressively increasing activity (95 sessions in total), the mean door-to-door times were consistently reduced to less than 9 minutes.