ESTRO 2020 Abstract book

S1083 ESTRO 2020

radiation oncologist and those obtained by the software from the old atlas, we obtained an overlap volume of 0.75 (SD = 0.11) and 0.75 (SD = 0, 17) respectively for the right and left breasts. In addition, twenty files (10 right and 10 left breasts) used to create the new atlas were tested to assess the automatic delineation of the mammary gland. The overlop volume in this case was 0.84 (SD = 0.09) for the right and 0.83 (SD = 0.08) for the left breasts. For OAR, after automatic delineation from the new atlas, the mean overlap volume was 0.86 (SD = 0.04) for the heart and 0.93 for each of the lungs (SD = 0.05 for the right lung and ET = 0.04 for the left lung). Conclusion Based on a new atlas, the Workflow box automatic delineation software has provided reliable OAR and mammary gland contours. The contours proposed by the software from the new atlas were better than those obtained from the previous atlas based on isodoses 95% of old treatment plans. The software has therefore become more efficient. Its daily use for patients with breast cancer treated in lateral position seems relevant. PO‐1850 Inter‐observer variation in delineating cardiac substructures and coronary arteries Z. Naimi 1 , R. Moujahed 1 , S. Ghorbel 1 , M. Ben Rejeb 1 , A. Hamdoun 1 , J. Yahyaoui 1 , L. Kochbati 1 1 Abderrahmen Mami Hospital, Radiation Oncology, Ariana, Tunisia Purpose or Objective The aim of the present study was to assess the inter‐ observer variation in delineating the cardiac chambers and coronary arteries as organ at risk in breast cancer radiotherapy. Material and Methods Three radiation oncologists delineated 11 cardiac substructures from CT planning images (without IV contrast) of 36 left‐sided breast cancer patients who received adjuvant 3D conformal RT. Cardiac substructures included great vessels, cardiac chambers, and coronary arteries (left main (LM), left anterior descending (LAD), right coronary artery (RCA) and circumflex (CX)). Delineation was performed according to the cardiac contouring atlas of F.Duane. Contouring variability was assessed using the DICE similarity coefficient (DSC) and the directed Hausdorff average distance (d H,avg ). Results The mean DSC for cardiac chambers ranged between 0.82 and 0.93 indicating considerable contour consistency. The similarity coefficients were 0.93, 0.92, 0.84 and 0.82 respectively for the left ventricle, right ventricle, left atrium and right atrium. The mean d H,avg for left ventricle was 1.07 mm and ranged between 0.68 and 1.5 mm for right ventricle. For coronary arteries, inter‐observer contour overlap (mean DSC) ranged between 0.49 and 0.67. Contour overlap was greater for the LAD than for the others coronary arteries with a mean DSC of 0.67. Despite poor similarity coefficients, inter‐observer contour separation was not meaningful as the mean d H,avg for the coronary arteries ranged between 1.3 and 2.7 mm. Conclusion Our study showed low inter‐observer variation in delineating cardiac substructures on CT simulation despite no contrast injection. The use of a cardiac contouring atlas enabled accurate and reproducible delineation of the heart chambers and coronary arteries. This would improve reliability of cardiac dosimetric studies. PO‐1851 Stereotactic radiation therapy for lung cancer: is a single 4‐dimensional scan enough? A. Gonzalez Moya 1 , P. Trémolières 1 , A. Paumier 1 , M. Mège 1 , J. Blanchecotte 1 , T. Lizée 1 , F. Legouté 1 , D.B. Christelle 2 , A. Damien 2 , D. Stéphane 2 1 Integrated Center for oncology Paul Papin Centre in Angers, Radiation oncology, Angers, France ; 2 Integrated

This is the first completely non‐invasive therapy for cardiac arrhythmias and could reduce procedure times for patients from up to 8 hours down to around 45 minutes, without risk of general anaesthesia or invasive ablation procedures. Material and Methods Non‐invasive 3D electro‐anatomical mapping and diagnostic CT scanning of the heart were used to define the VT substrate by Cardiology and an ITV developed to incorporate the VT target using RayStation (RaySearch Laboratories AB, Sweden) treatment planning system. Organs at risk were defined including coronary arteries and great vessels. The ITV was then grown to incorporate cardiac and respiratory motion based on a 4D‐CT. A VMAT plan was then created, using 3 partial arc 10MV FFF beams, with 25Gy prescribed to 95% of Planning target volume (PTV) in a single fraction. Individual CBCT parameters were developed to optimise image quality in preparation for treatment delivery. Results Organ at risk doses were kept to within mandatory constraints and a total dose of 25 Gy was delivered in a single 45 minute procedure. Pre, mid and post treatment CBCT’s were acquired and showed that intrafraction motion was within1 ‐ 2mm throughout. Prior to the treatment the patient had incessant VT and ICD shocks. 10 days post treatment it was observed that VT episodes were brief and non‐sustained. Conclusion This collaborative approach, has enabled the implementation of non‐invasive cardiac radioablation for ventricular tachycardia in our centre. This innovative technique may provide a new treatment option for patients who have exhausted conventional treatments. PO‐1849 Evaluation of an automatic contouring software in breast cancer treated in lateral position R. Msika 1 , N. Tkatchenko 1 , M. Robilliard 1 , A. Fourquet 1 , Y. Kirova 1 1 Institute Curie, Radiotherapy Department, Paris, France Purpose or Objective Irradiation of the breast in lateral decubitus in patients treated for breast cancer is an effective technique, allowing excellent sparing of organs at risk (OAR) such as the heart and lung. In our hospital, many patients are treated in this position, which implies a significant medical time devoted to the delineation of these patients. The purpose of this study is to evaluate the possibilities of improving the performance of automatic delineation software Workflow Box (Mirada Medical, UK) in the delineation of the breast and OAR in patients treated with isocentric lateral decubitus (ILD) technique. Material and Methods Currently, the automatic delineation of the mammary gland by this software is based on an atlas created from isodoses to 95% of previously treated patients. Forty CT scanners in treatment position (20 right and 20 left) of breast cancer patients delineated by three radiation oncologists, specialized in treatment of breast cancer led to the creation of a new atlas in automatic delineation software. Then 20 CT scans (10 right and 10 left breasts) were delineated by the software based on this new atlas and by radiation oncologist (contours of reference). Finally, these same twenty files were delineated from the old atlas. Contour accuracy was evaluated with the overlap volume index and the standard deviations (SD). Results In the breast delineation, comparing the volumes of software based on the new atlas and those of the expert physician, the average overlop volume was 0.80 (SD = 0.09) for the right breasts and 0.81 (SD = 0.06) for the left breasts. Comparing the target volumes delineated by the