ESTRO 2021 Abstract Book

S1534

ESTRO 2021

Results An in silico model of heterogeneous and realistic tumour oxygenation with a resolution of 10 µm was constructed. The vascular architecture resulted in tumours consisting of an outer more oxygenated shell enclosing a poorly oxygenated hypoxic core. The simulated tumours vary in diameter between 0.6-1.5 cm. In the well-oxygenated volumes the fractal dimension was 2.53 ± 0.08, the microvascular density 3451 ± 2281 vessels/mm 3 , and the hypoxic fraction 4.0% ± 3.5%. In the hypoxic volumes FD = 2.33 ± 0.10, MVD = 349 ± 152 vessels/mm 3 , and HF = 51.1% ± 17.9%. An example of the results is shown in Figure 2 as 2D slices from the center of the geometry illustrating the distribution of the tumour oxygenation at different scales, the corresponding vasculature and a detailed line plot of the oxygen partial pressure along one diameter.

Conclusion A versatile in silico model for simulating various patterns of tumour vasculature, the oxygenation and its dynamics on the microscale is presented. The model could therefore be a resourceful testbed when implemented in a treatment planning system, allowing for radiotherapy simulations under different microenvironmental scenarios with the ultimate goal of improving the success of each radiotherapy treatment. PO-1806 A framework for evaluating in vitro effects of GRID irradiation D. Arous 1,2 , B. V. Håland 1 , M. Børsting 1 , N. F.J. Edin 1 , E. Malinen 1,2 1 University of Oslo, Department of Physics, Oslo, Norway; 2 Oslo University Hospital, Department of Medical Physics, Oslo, Norway Purpose or Objective To present a framework for evaluating in vitro effects of GRID irradiation, including automated pixel-by-pixel mapping of surviving colonies and radiation dose. Materials and Methods A549 lung cancer cells cultured in vitro were irradiated using 220 kV X-rays with an open field or through a tungsten GRID collimator with sequential 5 mm openings and 10 mm blocking. Delivered doses were d=2, 5, and 10 Gy. Typically 30 000 cells were seeded in each flask to ensure appropriate colony scoring over a wide range of survival levels. A novel approach using principal component-based watershed method for image segmentation was used to locate the centroid of surviving colonies in scanned images of the cell flasks. Gafchromic film dosimetry and Monte Carlo simulations were employed to map the dose distribution in the flasks. From this, the dose at each surviving colony centroid was obtained. From cell surviving fraction (SF) following open field irradiation, a and b in the linear-quadratic survival model SF = exp(-ad-bd 2 ) was estimated by linear regression. The predicted survival level in each pixel was then mapped together with observed levels in the GRID-irradiated flasks. By this, a direct spatial comparison of predicted and observed survival levels was obtained, facilitating a better appraisal of GRID effects in vitro. Results Gafchromic field dosimetry (Figure 1) and Monte Carlo simulations gave highly similar dose distributions, with a mean peak- to-valley dose ratio of about 9. Fitting SF to clonogenic survival data for open field irradiation gave a=0.074±0.027 Gy -1 and b=0.026±0.003 Gy -2 . The novel colony segmentation method proved to demarcate surviving colonies with high accuracy (Figure 1). Mapping the SF longitudinally along the cell flask gave a pattern qualitatively resembling the GRID collimator outline (Figure 2). GRID irradiation with 10 Gy caused the greatest deviation from the prediction model based on homogeneous irradiation (Figure 1). The mean relative difference between predicted and observed survival in the (peak ; valley) dose regions was (3 ; 3) %, (25 ; 5)%, and (125 ; 7) % for 2, 5 and 10 Gy, respectively.