Abstract Book

S105

ESTRO 37

tumor misalignment, and CBCT3 after treatment. During treatment a video-based 3D optical surface monitoring system with 3 ceiling-mounted camera pods was used to monitor the real-time motion of a region of interest. This motion-monitoring was performed in 6 dimensions, namely lateral, longitudinal, vertical, rotational, pitch and roll. A tolerance of 0.3 cm for linear directions and 0.5° degrees for rotational directions was set, in accordance with a PTV margin of 3mm. If the optical surface monitoring system detected a violation of the set tolerance, treatment automatically stopped. If deemed necessary, the treating physician decided to acquire an extra CBCT for setup verification. Intrafraction motion was derived by registering CBCT2 and CBCT3 to the planning CT scan. A registration method in two phases was used. First, the bony anatomy of the CBCT was registered to the planning CT followed by registration of the tumor. The intrafractional shifts were calculated and compared with the set tolerances of the optical surface monitoring system (OSMS). For the shifts in each direction, mean, median and standard deviation were calculated. Results From June 2017 to September 2017, the records of 17 patients and 54 fractions were analysed. In 5 sessions treatment stopped because of violation of the set OSMS tolerance and the treating physician decided to acquire an extra CBCT for setup verification. For these sessions, the linear or rotational tumor shift detected on the extra CBCT was larger than the set tolerance of the OSMS, confirming the accuracy of the OSMS for detecting intrafractional tumor movement.

based on 1.5T Aera MRI). A volumetric T1w MR images (3D SPACE, TR/TE=420/7.2ms, voxel size=1.4x1.4x1.4mm 3 , scan time = 1.5mins) was then performed. The procedures were repeated for 10-20 times for each volunteer and a total of 185 sets of images were obtained. The 1 st MRI of each volunteer was used as a reference. By rigid image registration of 1 st MRI and the other MRI, shifts were recorded and setup reproducibility was investigated in terms of systematic, random error and 3-dimension vector length (3DVL). The results were compared with a previous clinical reproducibility study which was done in Tomotherapy machine using Mega Voltage Computed Tomography (MVCT) for position verification. Unpaired t-test and F test were used to compare the results from the two studies. Shifts of internal structures in the repeated setup were also investigated in this study. Seventeen head and neck structures were contoured in all the acquired MRI images. Shifts of center of mass for each structure were calculated. Results Calculated systematic errors, random errors, 3DVL and the previous study result were demonstrated in table 1. All 6D (Lateral, longitudinal, vertical, roll, pitch, yaw) systematic errors were very close to zero and neglectable. Random errors were also small and all below 1 mm except in longitudinal direction which had 1.05mm error. In statistical tests with previous study result, all compared errors were significantly smaller in MRI simulator study (p<0.05), except for lateral systematic error (p= 0.867). The structures shift in the repeated setup procedures in terms of systematic error, random error and 3DVL mean were shown in table 2. The head structures (brainstem, eyes, mandible, C1 and C2 vertebral body) were more stable than the neck and shoulder structures (C3 to T3 vertebral body, head of humerus). The reproducibility of the structures at head region was good (3DVL mean < 2mm, random error in all directions < 1.5mm) in MR simulator environment. However, for structures in the neck and shoulder region, the position shifts in the repeated setup were relatively large, especially for the heads of humerus (3DVL mean up to 4.3mm, random error up to 4.1mm in longitudinal direction).

For all the other sessions, the linear and rotational shifts detected on CBCT were within the tolerance level of the OSMS, except for one session were the difference in one rotational direction is 0.6°.

This results in an accuracy of the OSMS of 99.7%. Conclusion Our evaluation showed that real-time surface imaging provides sufficient accuracy for monitoring intrafraction motion in frameless SABR of nodal, spine and non-spine bone metastases. OC-0195 Reproducibility of head and neck setup in MR Simulator environment K.F. Cheng 1 , P.H. Fok 1 , G. Chiu 1 1 Hong Kong Sanatorium & Hospital, Division of Radiotherapy, Happy Valley, Hong Kong SAR China Purpose or Objective To investigate the reproducibility of head and neck setup in MR Simulator environment. Material and Methods Fifteen volunteers were recruited. Each volunteer was immobilized by head and neck thermoplastic cast. They were firstly setup and positioned with two Flex- 4 coils and one Body-18 coil covering the head and neck regions in MRI-Simulator (Siemens MAGNETOM RT Pro edition

Conclusion The reproducibility of head and neck setup in MR Simulator environment is good and at least no worse than the clinical result in tomotherapy treatment room.