8th Conference of the

Hungarian Association for Image Processing and Pattern Recognition

January 25-28, 2011
Szeged, Hungary

Gábor Székely

Computer Vision Laboratory
Department of Information Technology and Electrical Engineering, ETH Zurich
Zurich, Switzerland

Statistical models for supporting medical diagnosis, therapy and training

Statistical shape and appearance models have been oroginally proposed for supporting image segmentation based on prior knowledge, gained from manually segmented training examples. While these techniques are still very powerful tools for the segmentation of anatomical structures from two and three dimensional radiological images, a wide range of new possible directions for their clinical application have been identified during the past 15 years. The talk will present several examples, illustrating the potential of statistical models for supporting diagnosis, therapy planning, image guided navigation and medical education. Special emphasis will be given to their extension to describe spatio-temporal processes like organ motion and the investigation of their predictive power, allowing to predict information (like organ shape or appearance) from sparse and incomplete observations.

Biography: Gábor Székely studied chemical engineering and applied mathematics in Budapest, Hungary. He got his PhD in 1985 in analytical chemistry from the Technical University of Budapest. Between 1986 and 1990 he worked as a software specialist in a company producing NMR-equipment (Spectrospin, Fällenden, CH). In 1991 he joined the Computer Vision Laboratory of the ETH Zurich, where he got the venia legendi at the Department of Electrical Engineering in 1997. He has become Associate Professor in 2002 and full Professor in 2008 for Medical Image Analysis and Visualization. Since 2001 he is director of the Swiss National Center of Competence in Research on Computer Aided and Image Guided Medical Interventions (NCCR Co-Me, http://co-me.ch). His research concentrates on the development of information technology support tools for clinical diagnosis, therapy and training, as well as the application of simulation techniques for better understanding of complex biomedical processes.

Gabriela Csurka

Textual and Visual Pattern Analysis
Xerox Research Center Europe
Meylan, France

Multi-modal Fusion Strategies for Image Auto-annotation and Retrieval

Information, especially digital information, is no longer mono-modal: web pages can contain text, images, animations, sound and video; the valuable content within a photo sharing site can be found in tags and comments as much as in the actual visual content it contains. Nowadays, it is difficult to visit a page within a popular social network without finding a large variety of content modes surrounded by a rich structure of social information such as profiles, interest groups, consumer behaviors or simple conversations. The same situation holds for corporate documents and document collections. This major shift in the way we access content and what type of content we access poses a strong need for tools that enable interaction with multi-modal information.

The scientific challenge is therefore to understand the nature of the interaction between these modalities, and in particular how can we bridge the semantic gap between low-level features (extracted from images) and high-level concepts (in which users are interested on). Therefore, in the first part of the talk I will first present the Bag of Visual Words and Fisher Vectors image representations. These representations have shown state-of-the-art performance both in supervised (categorization) and unsupervised (retrieval) tasks. Then, after a brief introduction on text representation with Language Models, I will present different information fusion techniques (early and late fusion, cross-media similarity measures). I will show that if these image features are appropriately combined with textual information, the multi-modal system generally outperforms the mono-modal systems. I will finish the talk showing some of the applications of these multi-modal fusion strategies, such as image-auto annotation, multi-modal retrieval and content creation.

Biography: Gabriela Csurka is a research scientist in the Textual and Visual Pattern Analysis team at Xerox Research Centre Europe (XRCE). She obtained her Ph.D. degree (1996) in Computer Science from University of Nice Sophia - Antipolis. Before joining XRCE in 2002, she worked in fields such as stereo vision and projective reconstruction at INRIA (Sophia Antipolis, Rhone Alpes and IRISA) and image and video watermarking at University of Geneva and Institute Eurécom, Sophia Antipolis. Author of several publications in main journals and international conferences, she is also an active reviewer both for international journals (IJCV, TPAMI, TIP, PATREC, etc) and main computer vision conferences. Her current research interest concerns the exploration of new technologies for image content and aesthetic analysis, mono and cross-modal image categorization, semantic segmentation and retrieval.

Lilla Zöllei

A.A.Martinos Center for Biomedical Imaging
Massachusetts General Hospital
Charlestown, MA, USA

Robust and Accurate Registration of Structural MRI Images, Ex-vivo to In-vivo Acquisitions and Diffusion Tractography

1. G.M. Postelnicu, L. Zöllei, B. Fischl: "Combined Volumetric and Surface Registration", IEEE Transactions on Medical Imaging (TMI), Vol 28 (4), April 2009, p. 508-522
2. L. Zöllei, A. Stevens, K. Huber, S. Kakunoori, B. Fischl: "Improved Tractography Alignment Using Combined Volumetric and Surface Registration", NeuroImage 51 (2010), 206-213
3. L. Zöllei, A. Stevens, K. Huber, S. Kakunoori, B. Fischl: "Non-linear registration of intra-subject ex-vivo and in-vivo brain acquisitions" Human Brain Mapping, June 2010

Biography: Lilla Zollei is currently an Instructor at the Martinos Center for Biomedical Imaging, MAssachusetts General Hospital. Her research interests include pair-wise and group-wise image registration, information theoretical solutions in signal and image processing, diffusion imaging and the construction of digital brain atlases. She was born in Szeged, Hungary. She obtained her BA in Mathematics and Computer Science at Mount Holyoke College, South Hadley, MA and her MS and PhD in Electrical Engineering and Computer Science from the Massachusetts Institute of Technology, Cambridge, MA in 2002 and 2006, respectively. After finishing, she was awarded the Chateaubriand Fellowship and spent a year as a postdoc in the Applied Mathematics Department of Ecole Centrale de Paris, Paris, France with Prof Nikos Paragios. She then returned to Boston and started working with Bruce Fischl at the Laboratory of Computational Neuroscience, MGH first as a Research Fellow and then as an Instructor.

* The participation of the invited speakers is supported by:

European Union and the European Regional Development Fund, project TÁMOP-4.2.1/B-09/1/KONV-2010-0005.