ESTRO 38 Abstract book

S1152 ESTRO 38

occurrence of setup time, mathematically the region of median ± 34%. Hypoexponential distribution (red area) offered a much closer agreement to the observed setup time range (purple area) in all cases, in contrast to the normal distribution (green area) which has a much wider spread.

T.Y. Lee 1 , M.W.K. Law 1 , V.C.W. Cheung 1 , K.K. Tang 1 , S.W.K. Chow 2 , L.W.M. Lee 2 , M.M.S. Wong 2 , S.H. Yoo 1 , P.P.H. Nam 1 , J.W.Y. Lee 2 , B.S.K. Yu 1 1 Hong Kong Sanatorium & Hospital, Medical Physics and Research Department, Happy Valley, Hong Kong SAR China ; 2 Hong Kong Sanatorium & Hospital, Department of Radiotherapy, Happy Valley, Hong Kong SAR China Purpose or Objective The patient setup time was analyzed to provide a statistical insight regarding the variation of setup time, including the most probable setup time and spread across treatments of different anatomical regions. The major benefit of the study is to facilitate treatment time slot allocation, so as to improve treatment machine throughput without increasing the patient waiting time. Material and Methods The statistics of 3069 fractions out of 189 treatments, treated with 6MV photon-beam Volumetric-Modulated Arc Therapy (VMAT) by a linear accelerator (VersaHD, Elekta, Stockholm), were collected over a 25-month period. The setup time were inspected according to three treatment target regions: thorax (THX), breast (BRT) and head and neck (HN) as shown in Table 1. All first treatment fractions were excluded due to the high uncertainty in consultation and coaching time. The Hypoexponential distribution (Eq.1.) was fitted to the setup time of each of the three concerned treatment regions based on maximum log likelihood. Results The observed setup time histograms were well-followed by the fitted distributions (Fig. 1). The probability distributions of the overall observed setup time were positively skewed with tails extending to longer setup time. BRT had a wider spread and appeared similar to a normal distribution, while HN was heavily skewed and asymmetrical. The distribution of THX was in between of the two. HN took the shortest time to setup, followed by THX then BRT, with medians of 4.18, 6.08 and 7.83 minutes respectively. The optimal value of K was empirically found to be 8 based on our data. The model parameters of the three studied regions were found dissimilar after fitting (Table 1).

Conclusion This study investigated the statistical distribution of the patients’ setup time in 3069 fractions of VMAT treatments, categorized based on three anatomical regions. The setup time was shown to be effectively modeled by Hypoexponential distributions, a continuous probability model for describing the total finishing time of multiple sequential tasks such as queuing and logistic procedures. We have also shown the dissimilarity among the distributions of the three treatment regions, suggesting a treatment-region-based categorization scheme was essential for effective model fitting, so as to anticipate the patient setup time. In order to achieve a seamless and efficient treatment scheduling scheme, a representative probability model will help allocating treatment time slot based on treatment regions, thus achieving higher machine throughput while reducing patient waiting time simultaneously. EP-2088 Upright open-source cone beam CT imaging for radiotherapy J. Korte 1 , N. Hardcastle 1,2 , S. Everitt 3,4,5 , T. Kron 1,4 1 Peter MacCallum Cancer Centre, Physical Sciences, Melbourne, Australia ; 2 University of Wollongong, Centre for Medical Radiation Physics, Wollongong, Australia ; 3 Peter MacCallum Cancer Centre, Radiation Therapy Services, Melbourne, Australia ; 4 The University of Melbourne, Sir Peter MacCallum Department of Oncology, Melbourne, Australia ; 5 Monash University, Department of Medical Imaging & Radiation Sciences, Melbourne, Australia Purpose or Objective Some patients receiving radiotherapy may benefit from treatment in the upright position due to inability to lay prone/supine and through improved tumour-organ at risk geometry. A major challenge of upright radiotherapy

The shaded area under the fitted Hypoexponential distributions (Fig.2) shows the most probable 68%