ESTRO 2021 Abstract Book

S385

ESTRO 2021

to limit excessive treatment times. System latencies were quantified by using the portal imager on Unity.

Results The system latencies of 175 ms (MLC tracking) and 70 ms (gating) were effectively eliminated by using a linear regression predictor. The treatment delivery times were 5.7 mins (static, track-only), 8.8 mins (track+gate), and 20 mins (dual-gate). Dose difference maps confirm that the track+gate scenario best mimics the static dose distribution (Fig. 2). Dose profiles on the 2 (6.25) Gy isolines along the CC direction showed widening of 3.9 (-3.2) mm for track-only , 2.5 (-2.7) mm for dual-gate, and 0.5 (0) mm for track+gate deliveries with respect to the static reference. The gamma pass rates (2%/2mm) were 96.6% (track-only), 96.7% (dual-gate) and 99.4% (track+gate).

Conclusion This is the first experimental demonstration of simultaneous cardiorespiratory motion management for MRI- guided STAR resulting in very high dosimetric accuracy.

OC-0503 Characterisation of cardiac and respiratory motion in cone-beam CT images for cardiac SABR J. Daniel 1 , K. Burke 1 , P. Whitehurst 2 1 South Tees Hospitals NHS Foundation Trust, Medical Physics, Middlesbrough, United Kingdom; 2 The Christie NHS Foundation Trust, Medical Physics, Manchester, United Kingdom Purpose or Objective Cardiac SABR is a new treatment technique for ventricular tachycardia which delivers 25 Gy in a single fraction to the errant substrate in the myocardium. Planning challenges include the motion of both the diaphragm and the heart. We propose a method to track the heart and diaphragm in order to characterise motion for radiotherapy treatment planning and to quality assure patient set-up using raw CBCT projection data. Materials and Methods Planar images acquired as part of routine 4D Elekta XVI cone beam CTs included views of the heart and diaphragm over a range of 200°. The distal electrode (pacing lead tip) of implantable cardioverter devices (ICDs) were visible in the images and provided a suitable surrogate for target motion. Code written in Python 3 was used to process the images and identify the position of the pacing lead tip and level of the apex of the