ESTRO 2021 Abstract Book

S1394

ESTRO 2021

Finally, the feasibility of patient positioning in the linac room using CBCT-MR matching was verified, comparing the manual evaluations of three radiation oncologists with the shifts obtained through a standard CT-CBCT registration. Results Regarding geometrical distortions, the differences between the calculated and the nominal distances were always less than 2mm for each view. The mean deviation of DVH-parameters for PTV and OARs were lower than 1%. 2D-local gamma analysis, using an acceptance criterion of 2% and 2mm, provided average passing rates between 98.5% and 99.4% for all investigated planes. The averaged differences between CBCT-MR and the standard CT-based method for patient positioning resulted lower than 3mm for all directions. Conclusion This study demonstrates that MRI-only planning for prostate patients is feasible, providing a better contrast in the structures of clinical interest. In particular, sCT-based treatment plans provide dosimetric values in close agreement to original CT-based. Therefore, the future aim is to move towards an MRI-only patient approach in the clinical routine of prostate radiotherapy. PO-1671 Comparison between DIR of Eclipse and Velocity in PET/CT studies S. Pena Vaquero 1 , L. Cardoso Rubio 2 , L. Gómez Heras 2 , Á. del Castillo Belmonte 1 , M. Martín Veganzones 1 , C. Andrés Rodríguez 1 , M.M. Agulla Otero 1 , I. Conles Picos 1 , D. Miguel Pérez 1 , D. Alonso Hernández 1 , R. Torres Cabrera 1 1 Hospital Clínico Universitario, Servicio de Radiofísica y Protección Radiológica, Valladolid, Spain; 2 Hospital Clínico Universitario, Servicio de Oncología Radioterápica, Valladolid, Spain Purpose or Objective Nowadays, medical imaging plays an essential role in radiation therapy. Sometimes, the information provided by the planning TC can be limited concerning the contour process making necessary to rely on complementary studies such as PET or MRI. It is common for these studies to come from other centers or to be part of the set of images of the patient’s history, with no link to the radiotherapy process. It is in this context that image deformation tools acquire great importance. The aim of this work is to compare two Deformable Image Registration (DIR) algorithms using segmented structures by uptake threshold in lung PET/TC studies. Materials and Methods Ten patients treated for non-small-cell lung carcinoma with a single lung lesion and diagnostic PET/TC are retrospectively selected. The software for deformable fusion are Eclipse v15.6 (accelerated Demons) and Velocity v4.1 (extended multipass B-Spline). Structures are automatically generated in the deformed PET for each algorithm using a common relative threshold of the maximum SUV. To compare them, the Sørensen-Dice coefficient (DSC) and the relative variation of the generated volumes of each with respect to the union of the two are used.