Abstract book - ESTRO meets Asia

S121 ESTRO meets Asia 2018

Material and Methods 16 patients (2 adults, 14 children) treated with 3D-CRT for craniospinal irradiation were retrospectively chosen for this analysis. These patients were planned and treated during 2016-2017. Contouring of Brain and Spine Cord and organ at risk were already done and planning done on Eclipse TM Treatment Planning System (Varian). All of these patients were planned Lateral cranio-cervical fields and posterior spinal fields were planned using a forward- planned, FIF technique. Field junctions were automatically modulated and custom-weighted for maximal homogeneity within each treatment fraction. Dose volume histogram (DVH) was used for analysis of results. A corresponding plan without FIF technique was planned and maximum dose at the junction was noted for each patient with both plans and the readings were evaluated. Results For the FIF analysis, we studied 16 patients. Between 7 and 10 reduction fields were used. Field-in-field was primarily used to improve conformality of dose to the thecal sac, but plans were also evaluated on a single- fraction basis to judge the variation in dose delivered to the junctions each day. Without FIF or modulated junctions, the thecal sac volume receiving 110% of the prescribed dose ranged from 12% to 25.8% (mean: 15%) and volume receiving 120% dose ranged from 0% to 8% (mean: 4%). Whereas with FIF and modulated junctions, the thecal sac volume receiving 110% of dose ranged from 0.59% to 3% and volume receiving 120% was 0%. Conclusion Therefore it was concluded that Field-in-field technique improved dose conformality and reduced inhomogeneity. PO-295 Semantic segmentation using deep convolutional neural network in head and neck cancer radiotherapy K. Matsumoto 1 , T. Magome 1 1 Komazawa University, Department of Health Sciences, Tokyo, Japan Purpose or Objective Treatment planning for head and neck cancer radiotherapy requires accurate delineation of many structures. Manual contouring needs a long time and causes inter- and intra-observer variability. In recent years, semantic segmentation, which is a technology for recognizing and understanding in images at pixel level and classifying them, has been actively studied. The purpose of this study was to develop an automatic contouring technique based on the semantic segmentation using deep convolutional neural network (DCNN) for treatment planning CT images of head and neck cancer patients. Material and Methods Treatment planning CT images and structure data of 199 head and neck cancer patients in public database (the cancer imaging archive-head and neck squamous cell carcinoma) were used for this study. Label images were created with 9 structures (cord, brainstem, left / right parotid, etc.) and the remaining regions which were defined as “other” label. Two- dimensional treatment planning CT images were input to DCNN, and the label images were used as teacher data. In this study, encoder- decoder architecture called SegNet was tested as the DCNN model. The SegNet consists of the encoder network with 13 convolutional layers, the corresponding decoder network followed by a soft-max classifier for pixel-wise classification. In total 199 cases (30,690 slices), 80% of the cases (159 cases, 24,366 slices) were used for the training, and the remaining 20% cases (40 cases, 6,324 slices) were used for the evaluation of segmentation. The overlap between extracted regions and ground truth regions were evaluated by Dice coefficient.

C. Ma 1 1 Shandong Cancer Hospital, radiotherpy oncology physics, Jinan, China Purpose or Objective Texture analysis plays an important role in the research of PET image analysis. There are many studies investigating the applications of texture features in the clinical practice. The purpose of the study is to explore: 18F- Fluorodeoxyglucose (18F-FDG) uptake heterogeneity of primary tumor and lymphoma tumor by texture features of positron emission tomography (PET) image and quantify the heterogeneity difference between primary tumor and lymphoma tumor of non-small cell lung cancer (NSCLC) patients. Material and Methods Fourteen patients with primary tumors and lymphoma tumors of non-small cell lung cancer were enrolled in this study. All patients underwent whole-body 18F-FDG PET/CT scans. Firstly, texture features based on gray- level co-occurrence matrix (GLCM) were extracted to quantify the 18F-FDG uptake heterogeneity. Secondly, the Mann-Whitney U test was applied to texture features of primary tumors and those of lymphoma tumors. The values of texture features of primary tumors were compared with those of lymphoma tumors. Finally, the relationships between tumor volumes and texture features were analyzed. Results The values of three texture features including contrast, entropy, and correlation of lymphoma tumor were lower than those of primary tumor. However, the values of other three texture features including energy, homogeneity and inverse difference moment of lymphoma were higher than those of primary tumor. There are four texture features including energy, contrast, entropy and IDM have significant differences between primary tumor and lymphoma tumor. Furthermore, negative correlations between volume and three features including energy, homogeneity and inverse difference moment were observed. The study also found positive correlations between volume and the other texture features including entropy, contrast and correlation. Conclusion This study shows there are effective differences of heterogeneity between primary tumor and lymphoma tumor of non-small cell lung cancer by image texture analysis. PO-294 FiF Technique with Intrafractionally Modulated Junction shi fts for CSI Planning with 3D-CRT S.H. A. Ali 1 , H. Nazim 2 , R. Gohar 3 , J. Mallick 2 1 Ziauddin University, Radiation Oncology, Karachi, Pakistan 2 Ziauddin University Hospital, Radiation Oncology, Karachi, Pakistan 3 Aga Khan University Hospital, Radiation Oncology, Nairobi, Kenya Purpose or Objective To plan craniospinal irradiation with ‘‘field-in-field’’ (FIF) homogenization technique in combination with daily, intra-fractional modulation of the field junctions, to minimize the possibility of spinal cord overdose. Photon- based techniques for craniospinal irradiation (CSI) may result in dose inhomogeneity within the treatment volume and usually require a weekly manual shift of the field junctions to minimize the possibility of spinal cord overdose. Nowadays field-in-field technique is used to feather out the dose inhomogeneity caused by multiple fields. We have started using this technique after acquiring advanced technology machines in recent years.