ESTRO 36 Abstract Book

S95 ESTRO 36 _______________________________________________________________________________________________

correlation of visual acuity loss with the mean (r = 0.49, p = 0.001) and maximum (r = 0.47, p = 0.001) retina dose and tumor basal diameter (r = 0.50, p < 0.001). The dose to the macula showed no correlation with visual outcome (r = 0.24, p = 0.12). In the subgroup of patients with anterior tumor locations the maximum retina dose remained the only predictive factor (r = 0.46, p = 0.043). Evaluating the Cox proportional hazards model yielded a significantly higher risk for visual acuity loss (of more than 0.3 Snellen) for patients receiving a maximum dose of 500 Gy or higher (p = 0.009). A Cox multivariate analysis including the macula dose (p = 0.11) and basal diameter (p = 0.78) showed that a high maximum retinal dose is the highest risk factor (p = 0.017). The evaluation of the BED metrics showed no better correlation with the investigated endpoints and in some cases BED was even inferior. Conclusion The study showed that retina dose (D 2 and D mean ) is a suitable predictor for visual acuity loss, especially in case of anterior tumors where other risk factors (i.e. basal diameter) fail. References [1] R.G. Dale and B. Jones. The clinical radiobiology of brachytherapy. Br. J. Radiol. 71 , 465-483 (1998) PV-0186 MaxiCalc: a tool to calculate dose distributions from measured source positions in HDR brachytherapy M. Hanlon 1 , R.L. Smith 2 , R.D. Franich 1 1 RMIT University, School of Science, Melbourne, Australia 2 The Alfred Hospital, Alfred Health Radiation Oncology, Melbourne, Australia Purpose or Objective Dosimetric treatment verification via source tracking in HDR brachytherapy requires evaluation of the delivered dose as source dwell positions are detected. Current TPSs are not configured to perform this function, hence a fast dose calculation engine (DCE) that can accept the input of arbitrary dwell positions from the source tracking system is required. Here we present a TG-43 based DCE that computes 3D dose grids for measured dwell positions and performs a comparison with the treatment plan. Material and Methods The DCE, dubbed MaxiCalc, takes the input of measured dwell positions and times and calculates a dose grid of nominated dimensions and grid spacing for direct comparison to the treatment plan. MaxiCalc was validated against Oncentra Brachy (OCB v4.3) at 27 single dose points, as per OCB commissioning, as well as a 3D dose grid of 13 dwells. Dwell positions and times delivered in a phantom were measured by our source tracking system, as previously published. 1 The measured dwell positions were then used as input to MaxiCalc and the resultant dose grid compared to that from OCB. Observed dose differences due to source position measurement uncertainties were investigated. Results For the 27 dose points, MaxiCalc differed from OCB by a mean of 0.08% (σ=0.07%, max 0.41%) demonstrating differences that are similar to those between published values 2 and OCB. In a multi-source plan for doses between 50-200% of the prescription dose, MaxiCalc yields a maximum difference of <1%, which arises due to minor calculation differences in the steep dose gradients near the source. There was a gamma pass rate of >99% at 1mm/1%. A dose grid was calculated for a plan of 25 dwell positions acquired using our source tracking system, there was maximum difference of 12.2% (mean = 0.7%). The maximum difference arises from a small shift in the apparent dwell positions causing large differences due to the high dose gradients near the source, which is only significant within 10 mm of the source. For this volume of

interest, only 0.2% of voxels differ by >5%, showing good agreement throughout. Results from measured delivery errors, such as those in figure 1, will also be presented.

Figure 1: Dosimetric comparison between planned and delivered doses for a HDR brachytherapy treatment in a phantom with an introduced error. Conclusion Real-time dosimetric treatment verification is possible with our source tracking system combined with MaxiCalc. Fast dose calculation based on measured source dwell positions is achieved and overcomes the limitation of current TPSs. References 1. Smith, R L., et al. Medical physics 43.5 (2016): 2435-2442. 2. Daskalov, G M., et al. Medical physics 25.11 (1998): 2200-2208. PV-0187 Source dwell time and transit time measurement for a HDR afterloading unit T.L. Chiu 1 , B. Yang 1 , H. Geng 1 , W.W. Lam 1 , C.W. Kong 1 , K.Y. Cheung 1 , S.K. Yu 1 1 Hong Kong Sanatorium & Hospital, Medical Physics & Research Department, Happy Valley, Hong Kong SAR China Purpose or Objective To evaluate dwell time and transit time of HDR brachytherapy treatment by an in-house fluorescent screen based QA system. Since dosimetric effect would be directly affected by source dwell time, an accurate QA method on temporal accuracy is essential. Material and Methods The system included a fluorescent screen (Kodak, Lanex regular screen) which converts the radiation signal to optical signal and a high-speed camera with frame rate up to 500 fps and pixel resolution of 1280X720. The temporal resolution was 2 ms. A catheter in which an Ir-192 source would be loaded was fixed on the fluorescent screen and the camera was placed 30 cm away from the screen. The whole system was light-shielded. When the source travelled inside the catheter, the camera would capture images on the fluorescent screen sequentially. Source position was traced out by locating the centroid of the captured image. The accuracy of dwell time was assessed by measuring 3 different dwell times, namely, 1 s, 0.5 s & 0.1 s. According to a white paper from vendor, transit time for separations below 35 mm would occupied part of the next dwell time and those for separations above 35 mm would have 0.1 s compensation. Thus, the influence of transit time on dwell time was studied by measuring 0.5 s dwell time under 3 different dwell separations, namely, 6 cm, 4 cm & 0.5 cm. Dwell time was assessed by counting the number of images in which source positions were unchanged to the subsequent image. Transit time was the time between two dwell positions. Results