ESTRO 2020 Abstract book

S971 ESTRO 2020

separation

Conclusion In this study a difference between types of reconstruction was observed. The significant difference was observed only for high doses. The biggest differences were observed for calculation of dose performed on standard CBCT reconstruction images. Dose calculation performed on images from CT and iCBCT was very similar. The iCBCT algorithm has great ability to reduce noise and metal artifact and can be used for adaptive radiotherapy routinely and for patients with metal implants. PO-1673 Head-and-neck tumor delineation for MRI guided adaptive (chemo)radiotherapy M. Philippens 1 , B. Peltenburg 1 , P. Doornaert 1 , F. Pameijer 2 , A. Kotte 1 , R. De Bree 3 , C. Terhaard 1 1 UMC Utrecht, Department of Radiation Oncology, Utrecht, The Netherlands ; 2 UMC Utrecht, Department of Radiology, Utrecht, The Netherlands ; 3 UMC Utrecht, Head and Naclk Oncological Surgery, Utrecht, The Netherlands Purpose or Objective Patients with head and neck squamous cell carcinoma (HNSCC) treated with radiotherapy can potentially benefit from online adaptive radiotherapy planning using MRI. However, changes in tumor visibility on MRI during therapy may hamper the contouring process and decrease the reliability of adaptive planning. As the ground truth is missing, interobserver agreement is used as a measure for precision. This study aimed at determining the change in interobserver agreement of the delineation of head and neck tumors on T2 weighted MRI during Twenty patients with HNSCC treated with (chemo)radiotherapy received four MRI exams; one prior to treatment and three during the first weeks of treatment. Three observers delineated the tumor on the intratreatment T2 weighted MRI scans with mDixon fat (chemo)radiotherapy. Material and Methods

using clinical contouring guidelines for pretreatment tumor delineation. Interobserver agreement was determined using conformity index and by measuring the maximum distance between delineations. Results The conformity index decreased with each progressing week: 0.58 (week 2), 0.52 (week 3) and 0.42 (week 4). In addition, the confidence of the observer in the delineation ceased over the weeks. Average tumor volume delineated also decreased from 15.1 ml prior to treatment to 5.4 ml in week 4. Maximum distance between the delineations of the three observers remained constant at around 1.5 cm for all weeks. Conclusion Interobserver agreement decreased as radiotherapy progressed. For clinical introduction of adaptive radiotherapy based on MRI, guidelines and additional training are needed to improve the consistency of target definition.