Abstract Book

S1175

ESTRO 37

were more reproducible than either shape metrics or textural features, but with some dependence on the imaging modality. Entropy was consistently one of the most stable first-order features. There was no emergent consensus regarding either shape metrics or textural features, however coarseness and contrast consistently appeared among the least reproducible. The qualitative synthesis has been summarized in Figure 2. Conclusion Investigations of feature repeatability and reproducibility are currently limited to a small number of cancer types. Reporting quality may be improved in regards to details of: feature extraction software, digital image manipulation (pre-processing) and cut-off value used to distinguish stable features. We recommend that authors publish image collections (including delineations) and feature extraction software as open access resources, to accelerate progress in clinical generalizability and independent validation of models.

treatment outcome, however, manual delineation of the prostate is not only time consuming but requires considerable clinical experience and knowledge on how best to interpret multi-modality image information from computerised tomography (CT) and magnetic resonance (MR) image data. The aim of this work was to assist with this task and construct a prostate shape model, or atlas, that would automatically segment the prostate on MR images. Material and Methods T2-weighted MR prostate scans from 40 patients (30 training cases and 10 test cases) were obtained from the publicly available MICCAI PROMISE12 data set. Following registration of all prostate cases in the training set to a reference case an atlas was created using Principal Component Analysis (PCA) to extract the mean-shape and shape variations from the data. A covariance matrix was computed by subtracting the average shape and decomposing the matrix to find the first few large eigenvalues covering 95%~98% of the information within the shape. The shape model was then structured as: prior model = (mean shape + b x Ø), where b is the weight vector corresponding to the eigenvectors and Ø is the corresponding eigenvector matrix. Any new case can be generated by deforming the prior atlas. Results Table 1 shows the simulation performance using three atlas shape model methods incorporating different variations of the algorithm. These include rigid and non- rigid registration for the initial registration to a reference case and Iterative Closest Point (ICP) and Coherent Point Drift (CPD) to search for corresponding pairs in the registered data. The pilot results show that the CPD method works better than ICP registration, particularly for non-rigid data points. Furthermore, both the Dice Similarity Criterion (DSC) and the Jaccard Index (JI) measures for the CPD method indicate an acceptable performance with mean values of 0.71 and 0.55 for the combined approach. The Hausdorff Distance (HD), with a mean of 4.04 mm, also represents a favourable result for this method. Table 1 : Performance comparison of three different models used to build the atlas. Conclusion The proposed atlas-based segmentation algorithms were evaluated on T2-weighted MR prostate images from the MICCAI PROMISE12 challenge. The simulation results demonstrate that the proposed approach has generally acceptable performance for automatically segmenting a new prostate case and has the potential to be used as part of a clinical decision support system. However, further validation on a much larger data set is required. EP-2134 Texture features from Tomo daily MVCT may predict radiotherapy outcome of NSCLC: a pilot study J. Zhang 1 , J. Zhu 2 , Y. Zhang 1 , Y. Yin 2 , B. Li 3 1 University of Jinan, Shandong Provincial Key Laboratory of Network Based Intelligent Computing, Jinan, China 2 Shandong Cancer Hospital affiliated to Shandong University, Department of Radiation Oncology Physic and Technology, Jinan, China 3 Shandong Cancer Hospital affiliated to Shandong University, Department of Radiation Oncology, Jinan, China Metric Rigid-ICP Rigid- CPD Rigid- & Non- rigid-CPD Dice Similarity Criterion 0.625 0.659 0.706 Jaccard Index 0.456 0.491 0.546 Haussdorf Distance 18.040 4.518 4.041

EP-2133 Automatic Segmentation of the Prostate Using Shape Models T. Ding 1 , D. McLaren 2 , W. Nailon 3 1 The University of Edinburgh- Institute for Digital Communications, School of Engineering, Edinburgh, United Kingdom 2 Edinburgh Cancer Centre- Western General Hospital-, Department of Clinical Oncology, Edinburgh, United Kingdom 3 Edinburgh Cancer Centre- Western General Hospital-, Department of Oncology Physics, Edinburgh, United Kingdom Purpose or Objective In the UK, prostate cancer accounts for 25% of all new cancers in males with radiotherapy one of the main treatments for this disease. Accurate delineation of the prostate is therefore essential to ensure the best