Abstract Book

S152

ESTRO 37

assessing whether a cGAN can generate sCT suitable for treatment planning on prostate cancer patients. Material and Methods A study was conducted on 59 patients who underwent prostate IMRT for which CT (Brilliance CT Big Bore, Philips) as well as MR (3T Ingenia Omega HP, Philips) scans were acquired for simulation purposes on the same day and in RT position. To generate the sCT images, dual gradient-echo, RF spoiled, 3D T1w MR images were acquired with 1x1x2.5 mm3 resolution, TR/TE2/TE1=3.9/2.5/1.2ms. Dixon reconstruction was performed producing water (W), fat (F) and in-phase (IP) images. A cGAN called “pix2pix” (P. Isola et al, 2016, arXiv) was used in this experiment. Before training, CT images were rigidly registered to MR images and MR images were normalised (Figure 1a). Within pix2pix, the images were scaled to 8-bit grayscale values and resampled to 256x256-pixel slices (Figure 1b). Training of pix2pix was performed on 32 patients in 2D transverse slices using 200 epochs. To guarantee consistent air pockets on the CT and MR images during training, internal air pockets as detected on MR images were copied to CT (Figure 1c). The trained cGAN was applied to the remaining 27 patients (test set) producing sCT images. Image evaluation was performed on the sCT using mean absolute error (MAE) as compared to CT. Dose recalculation of clinical 5-beam 10 MV IMRT plans with a prescribed dose of 35x2.0 Gy to the target was performed for 15/27 patients on the CT and sCT images in Monaco (v 5.11.02, Elekta AB). Dose distributions were subsequently analysed through voxel-based dose differences and gamma analysis.

Conclusion Results suggest that accurate MR-based dose calculation using a 2D cGAN for sCT generation is feasible for prostate cancer patients. An additional advantage of using this network is the short time needed to generate the sCT, which would be beneficial for MR-guided RT application. Future investigations will evaluate dose to clinically relevant volumes as well as the use of a 3D network. OC-0295 The feasibility of volumetric 4DMRI in upper abdominal radiation therapy treatment planning A. Oar 1 , R. Rai 1 , M. Jameson 1 , S. Deshpande 1 , G. Liney 1 , E. Juresic 1 , J. Veneran 1 , G. Dinsdale 1 , D. Elwadia 1 , S. Kumar 1 , M. Lee 1 1 Liverpool Hospital, Department Radiation Oncology, Liverpool, Australia Purpose or Objective 4DCT is widely used in radiation therapy (RT). 4DCT uses additional radiation exposure and has limitations in soft tissue delineation. Although 2DMRI ( MRIcine ) has been utilised in tumour tracking, there is very little published data utilising 3D volumetric acquired 4DMRI to detect tumour motion. 4DMRI provides motion information with improved volumetric tumour definition and without additional radiation exposure. Here we explore the movement of tumour and surrogates of tumour position in five patients undergoing upper abdominal RT on 4DCT and volumetric 4DMRI. Also, we explore the feasibility of abdominal compression (AC) in three healthy volunteers undergoing 4DMRI. AC reduces respiratory amplitude and liver movement. AC in combination with volumetric 4DMRI may benefit patients having liver SBRT. Material and Methods Patients underwent standard RT simulation including 3D and 4D CT and MRI simulation in the treatment position. The 4DMRI sequence is a prototype T1 weighted 3D gradient echo (VIBE) with radial self-gating (Siemens, Erlangen, Germany). For each patient, tumour or a surrogate of tumour position was selected and the range of motion on 4DCT and 4DMRI was compared. Volumes were contoured (MIM Software Inc, Cleveland, USA) on maximal inspiratory and expiratory phases for both 4DCT and 4DMRI. The total distance between centroids was calculated using Euclidian distances. Ability to delineate tumour and image quality were graded as per following: tumour-edge or image very clear (1), tumour-edge or image slightly blurred (2), considerable blurring of tumour-edge or image (3) and unable to delineate tumour or image not useable (4). Furthermore, three healthy volunteers were scanned using 4DMRI, with and without AC to confirm the feasibility of the combination of AC

Results In total, 3495 slices were used for training, requiring about 11 hrs on a GPU Tesla P100 (NVIDIA). Applying the trained cGAN to a single patient volume (Figure 2) required about 5.6s. Image Evaluation: On average, an MAE of 65±10HU (±1σ, range: 50-96HU) was obtained in the intersection of the body contours between CT and sCT. When air pockets were also copied to the CT in the test set, the MAE reduced to 60±6HU (range: 48-71HU). Dose Evaluation: On average (Table 1), a dose difference below 1.1% was obtained using a 50% dose threshold of the prescribed dose. A mean gamma pass rate of 96±4% was obtained.