VISCERAL@ISBI 2015 VISCERAL Anatomy3 Organ Segmentation Challenge co-located with IEEE International Symposium on Biomedical Imaging 2015 New York, NY, USA, April 16, 2015 Proceedings Orcun Goksel Oscar Alfonso Jiménez del Toro Antonio Foncubierta Rodrı́guez Henning Müller (Eds.) c 2015 for the individual papers by the papers’ authors. Copying permitted only for private and academic purposes. Re-publication of material from this volume requires permission by the copyright owners. Corresponding editor’s address: Prof. Dr. Orcun Goksel ETH Zürich – Computer Vision Lab Sternwartstrasse 7 8092 Zürich, Switzerland ogoksel@ethz.ch Preface VISCERAL (Visual Concept Extraction Challenge in Radiology) aims to organize series of benchmarks on the processing of large-scale 3D radiology images, by using an innovative cloud-based evaluation approach. While a growing number of benchmark studies compare the performance of algorithms for automated organ segmentation in images with restricted field of views, emphasis on anatomical segmentation in images with wide field-of-view (e.g. showing entire abdomen, trunk, or the whole body) has been limited. VISCERAL Anatomy benchmark series aim to address this need by providing a common image and test dataset and corresponding segmentation challenges for a wide range of anatomical structures and image modalities. This proceedings summarize the techniques submitted for Anatomy3 benchmark, the results of which were also presented at the ISBI VISCERAL Challenge session on April 16th 2014, as part of the IEEE International Symposium on Biomedical Imaging (ISBI) in New York, NY, USA. The challenge participants used an online evaluation system, where they submitted their algorithms in a virtual machine environment. The organisers then run the virtual machines on the test images and populated the segmentation results in a participant viewable results board. Then, the participants could at their discretion upload their results to a public leaderboard. The results from the methods presented here were published in the online leaderboard two weeks before the challenge session. The short papers in this proceedings were submitted by the participants to describe their specific methodologies used to generate their results. At the session, participants had a chance to present their methods as oral presentations. We thank the authors for their submissions and the program committee for their hard work. Proceeding editors On behalf of VISCERAL Consortium 3 Session Chair Orçun Göksel, ETH Zürich, Switzerland Proceedings Editors Orçun Göksel, ETH Zürich, Switzerland Oscar Jiménez, University of Applied Sciences Western Switzerland, Switzerland Antonio Foncubierta, ETH Zürich, Switzerland Henning Müller, University of Applied Sciences Western Switzerland, Switzerland VISCERAL Consortium Allan Hanbury, Vienna University of Technology, Austria (coordinator) Henning Müller, University of Applied Sciences Western Switzerland, Switzerland Georg Langs, Medical University of Vienna, Austria Orçun Göksel, ETH Zürich, Switzerland Bjoern Menze, Munich University of Technology, Germany Marc-André Weber, University of Heidelberg, Germany Tomàs Salas Fernandez, Catalan Agency for Health Information, Assessment and Quality, Spain 4 Contents PART I: Organization and Evaluation Overview of the VISCERAL Challenge at ISBI 2015 Orcun Goksel, Antonio Foncubierta–Rodrı́guez, Oscar Alfonso Jiménez del Toro, Henning Müller, Georg Langs, Marc-André Weber, Bjoern Menze, Ivan Eggel, Katharina Gruenberg, Marianne Winterstein, Markus Holzer, Markus Krenn, Georgios Kontokotsios, Sokratis Metallidis, Roger Schaer, Abdel Aziz Taha, András Jakab, Tomàs Salas Fernandez, Allan Hanbury 6 PART II: Participant Submissions Good Features for Reliable Registration in Multi-Atlas Segmentation Fredrik Kahl, Jennifer Alvén, Olof Enqvist, Frida Fejne, Johannes Ulé, Johan Fredriksson, Matilda Landgren, Viktor Larsson 12 Fully Automatic Multi-Organ Segmentation Based on Multi-Boost Learning and Statistical Shape Model Search Baochun He, Cheng Huang, Fucang Jia 18 Hierarchic Anatomical Structure Segmentation Guided by Spatial Correlations (AnatSeg-Gspac): VISCERAL Anatomy3 Oscar Alfonso Jiménez del Toro, Yashin Dicente Cid, Adrien Depeursinge, Henning Müller 22 Multi-modal Multi-Atlas Segmentation using Discrete Optimisation and Self- Similarities Mattias P. Heinrich, Oskar Maier, Heinz Handels 27 Efficient and fully automatic segmentation of the lungs in CT volumes Yashin Dicente Cid, Oscar Alfonso Jiménez del Toro, Adrien Depeursinge, Henning Müller 31 5