ESTRO 2021 Abstract Book

S576

ESTRO 2021

Conclusion Prostate cancer plans created using Personalized engine provided an overall increase of plan quality, in terms of dose conformity and sparing of normal tissues. The Feasibility “a priori” DVH prediction module provided OARs dose sparing well beyond the clinical objectives. The new Pinnacle Personalized algorithms outperformed the currently used Autoplanning ones as solution for treatment planning automation. PD-0747 Knowledge-based approach for DVH prediction in robotic spine SBRT. S. BROGGI 1 , R. Castriconi 1 , A. Tudda 1 , C. Deantoni 2 , A. Fodor 3 , B. Longobardi 1 , L. Perna 1 , P. Mangili 4 , N.G. Di Muzio 3 , A. Del Vecchio 1 , C. Fiorino 1 1 IRCCS San Raffaele Scientific Institute, Medical Physics, Milano, Italy; 2 IRCCS San Raffaele Institute, Radiotherapy, Milano, Italy; 3 IRCCS San Raffaele Scientific Institute, Radiotherapy, Milano, Italy; 4 IRCCS San Raffaele Milano, Medical Physics, Milano, Italy Purpose or Objective Aim of this study was to evaluate the clinical feasibility of a Knowledge – based (KB) dose-volume histogram (DVH) prediction model for spine SBRT delivered with robotic radiosurgery. Materials and Methods Thirty three SBRT spinal metastases treated with SBRT (Cyberknife® M6, CK) were collected. All cases were referred and planned (Precision v. 2.0.1.1) based on the RTOG 0631 criteria; a single dose session of 18 Gy (28 pts) or 20 Gy (5 pts) was prescribed to PTV with a strict dose fall-off requirement for both point (D 0.03cc = 14 Gy) and, with slightly decreasing priority, volumetric (V10=0.35 cc, V7=1.2 cc) dose constraints for spinal cord, defined, when possible, on MRI-CT matched images. Prescribed dose was normalized at the peripheral isodose (median: 77%; range: 70-86%) with median V18/20Gy of PTV equal to 92.2 %. CT images, structures and 3D dose cubes were exported to Eclipse system (v. 13.6) and used to train a KB-model using the Varian Rapid Plan tool. The 3D dose cubes were linked to a volumetric arc planning (“fake-plan” consisting of two arcs 130°- 230°; 230°-130°) necessary to assess the association between the principal dosimetry components (PCA) of DVHs and anatomy/geometry components. PTV, spinal cord and planning spinal cord (PRV, 2 mm margin) were considered to generate the DVH prediction model. Results 32 plans were finally considered, excluding one dosimetric outlier. Final model resulted in a linear regression between dosimetry and anatomy PCA, with R 2 equal to 0.85 and 0.82 for spinal cord and PRV respectively. When comparing clinical vs predicted DVHs (once applied the model), the clinical DVHs were generally within the lower and upper band values estimated by the model: as an example in figure 1, V10 of the spinal cord of