ESTRO 36 Abstract Book

S875 ESTRO 36 _______________________________________________________________________________________________

References [1] Brouwers PJ et al. Set-up verification and 2- dimensional electronic portal imaging device dosimetry during breath hold compared with free breathing in breast cancer radiation therapy. Pract Radiat Oncol. 2015 May- Jun;5(3):e135-41 [2] Cerviño L et al. Using surface imaging and visual coaching to improve the reproducibility and stability of deep-inspiration breath hold for left-breast-cancer radiotherapy. Phys. Med. Biol. 54 (2009) 6853–6865 EP-1618 Can diaphragm motion function as a surrogate for motion of esophageal tumors during treatment? S.E. Heethuis 1 , L. Goense 1 , A.S. Borggreve 1 , P.S.N. Van Rossum 1 , R. Van Hillegersberg 2 , J.P. Ruurda 2 , S. Mook 1 , G.J. Meijer 1 , J.J.W. Lagendijk 1 , A.L.H.M.W. Van Lier 1 1 University Medical Center Utrecht, Department of Radiotherapy, Amsterdam, The Netherlands 2 University Medical Center Utrecht, Department of Surgery, Amsterdam, The Netherlands Purpose or Objective Esophageal tumors show large motion in cranio-caudal direction (CC), with a Peak-to-Peak (P-t-P) range of 2.7 to 24.5mm [Lever F. et al. (2013)]. In case the motion of the tumor could be followed during radiotherapy treatment, this would enable treatment margin reduction. The aim of this research is to investigate whether the motion of the diaphragm is correlated with breathing motion and drift we can detect in esophageal tumors. As such, the diaphragm could function as a surrogate for esophageal tumor motion during treatment. Material and Methods In total, 46 coronal cine MR scans were obtained from 4 patients whom were treated with neoadjuvant chemoradiotherapy (nCRT) for distal esophageal cancer. In this study, one MR scan was performed prior to nCRT, followed by 5 weekly MR scans during nCRT (in one patient only 4 scans). Cine MR scans included 75 frames acquired in approximately 45 seconds, with a resolution of 2.01x2.01mm. The scan was acquired twice within one session, separated by circa 10 minutes. To estimate motion in the cine MR series an optical flow algorithm (RealTITracker, [Zachiu C. et al. (2015)]) was used to calculate motion fields. The tumor was delineated manually, in which the mean motion for each frame was calculated in CC direction. Motion was also estimated in the diaphragm/liver border within a manually placed rectangle. An in-house tool was designed to find peaks and estimate drifts in the motion curves. Drift was defined as the change in the mean between consecutively found local maxima and minima. Correlation of the CC motion between diaphragm and tumor was calculated. P-t-P analysis was performed on tumor motion curves and tumor motion curves corrected for drift using the diaphragm drift A strong Pearson’s correlation of r=0.972 was found while comparing CC motion in diaphragm and tumor, with a range of 0.849-0.996. The mean P-t-P tumor motion before and after correction for drift was 10.1 and 9.3mm respectively (p<0.05). However, for individual scan sessions the effect of drift could be much larger, as is exemplified in Fig. 1a . P-t-P amplitude for each patient before and after drift correction is shown in Fig. 2 . Although the amplitude of the diaphragm motion was higher, mean P-t-P motion of 12.6mm, when the tumor motion showed a drift or sudden movement, this was also found in the diaphragm motion ( Fig. 1&2 ). Conclusion In this study it was found that diaphragm motion shows a strong correlation with esophageal tumor motion. Using the diaphragm motion for drift correction resulted on ( Fig. 1 ). Results

average in a reduction of the P-t-P range over all patients. This reduction can be used for adaptive treatment strategies, which reduce margins. For example, in case an MR-linac is taken in mind [Lagendijk J.J.W. et al (2008)], MR-based gating to compensate for respiratory motion and/or base-line shift (drifting) detection using the diaphragm as surrogate will be well feasible.

EP-1619 Determination of Lung Tumour Motion from PET Raw Data used for Accelerometer Based Motion Prediction G. Hürtgen 1 , S. Von Werder 2 , V. Berneking 1 , K. Gester 1 , O. Winz 3 , P. Hallen 4 , F. Büther 5 , C. Schubert 1 , N. Escobar-Corral 1 , J. Hatakeyama Zeidler 6 , H. Arenbeck 6 , C. Disselhorst-Klug 2 , A. Stahl 7 , M.J. Eble 1 1 RWTH Aachen University Hospital, Department of