Emmanuel Caruyer
Department of Radiology, University of Pennsylvania, USA

Maxime Descoteaux
Sherbrooke Connectivity Imaging Lab (SCIL), Universit´e de Sherbrooke, Canada

Jean-Philippe Thiran
Signal Processing Lab (LTS5), EPFL, Switzerland
University Hospital Center (CHUV) and University of Lausanne (UNIL), Switzerland

Website
http://hardi.epfl.ch/static/events/2013_ISBI/

Challenge Abstract
Validation is the bottleneck for the diffusion magnetic resonance imaging (MRI) community. Diffusion MRI is a powerful MRI modality which is sensitive to the random movement of the water molecules in biological tissues. By studying the anisotropy of this diffusion process in white matter it is possible to highlight structures otherwise invisible with other imaging modalities and to infer the neuronal wiring of the brain. The study of this connectivity is of major importance in a clinical perspective, with particular emphasis on neurological disorders which, nowadays, affect up to one billion people worldwide.

The state-of-the-art Diffusion Spectrum Imaging (DSI) [1] modality is known to provide good imaging quality but is too time-consuming to be of real interest for practical applications. Accelerated acquisitions, relying on a smaller number of sampling points, are thus required. In the last few years a multitude of new reconstruction approaches have been proposed to recover the local intra-voxel fiber structure. Some of them aim at improving the quality of the reconstructions while others focus on reducing the acquisition time. However, when a new algorithm is proposed, the performances are normally assessed with ad-hoc synthetic data and evaluation criteria. Hence, when proposing a new algorithm, it can be really difficult to compare the performances against the state-of-theart techniques. This aspect is even more crucial in a clinical perspective, as the availability of a comprehensive comparison of available reconstruction methods, highlighting strengths and weaknesses of each approach, might help clinicians in the choice of the most adequate diffusion MRI technique for a specific clinical application.

The diffusion MRI reconstruction challenge is organized with the aim to provide all researchers in this field with a common framework to assess the performances of their own algorithms and fairly compare their results against other approaches. The purpose of the contest is to compare different reconstruction schemes on the same synthetic data and under controlled conditions. The competitors will be asked to provide the best estimation of the configuration of fiber compartments of a common dataset by using their proposed algorithms.

Stamatis Lefkimmiatis,
Postdoctoral researcher, Biomedical Imaging Group,
Ecole Polytechnique Federale de Lausanne (EPFL) (Lausanne, Switzerland)

Laure Blanc-Feraud,
CNRS Research Director, Laboratoir I3S (Sophia-Antipolis, France)

Michael Unser,
Head of the Biomedical Imaging Group,
Ecole Polytechnique Federale de Lausanne (EPFL) (Lausanne, Switzerland)

Rainer Heintzmann,
Head of the Microscopy Reseach Unit, Institute of Photonic Technology (Jena,
Germany) and Nanobiophotonics Professor for Physical Chemistry,
Friedrich-Schiller-Universitat (Jena, Germany)

Arne Seitz,
Head of the Bioimaging and Optics Platform,
Ecole Polytechnique Federale de Lausanne (EPFL) (Lausanne, Switzerland)

Website
http://bigwww.epfl.ch/deconvolution/challenge/

Challenge Abstract
This document describes a project for a challenge in the framework of the 2013 ISBI conference. It is centered around 3D deconvolution in uorescence microscopy, which is one of the most common inverseproblem incarnations in bioimaging. Deconvolution also admits many variants with a one gradation in terms of complexity (e.g., blind deconvolution, space-variant image restoration or extension to super-resolution techniques such as structured illumination). We thus believe that the present project is suitable for creating a series of recurring annual challenges on the same topic.

The challenge will be organized as a three-stage tournament (training, qualification and final). Beyond selecting the most promising deconvolution algorithms and practitioners, our goal is to promote crossfertilization between the relatively independent communities of academic, open-source and commercial developers.

The evaluation of the results will be based on three tools: 1) a realistic 3D phantom of a mitotic cell that will serve as ground-truth data, 2) a physically motivated forward model and 3) a set of performance metrics chosen from three complementary categories (general-purpose, sample-specific and task-oriented metrics).

John Sedat is Professor Emeritus in Biochemistry and Biophysics at the University of California, San Francisco. He received his Ph.D. in Biology from Caltech in Pasadena, CA in 1970 followed by postdoctoral study under Fred Sanger at the MRC in Cambridge, England. Subsequently, he spent approximately a year at the Hebrew University at Hadassah in Jerusalem, Israel where the Drosophila and chromosome study started.  Several years at Yale followed, and he has been at UCSF since 1977.

His scientific efforts are directed toward deciphering higher order chromosome structure, an unsolved problem, using Drosophila as a model system. Interestingly, those studies have required development, in collaboration with his colleague, David Agard, of several optical microscope systems to adequately visualize and quantitate the intricate chromosomal organization.

Donoho is a mathematician who has made fundamental contributions to theoretical and computational statistics, as well as to signal processing and harmonic analysis. His algorithms have contributed significantly to our understanding of the maximum entropy principle, of the structure of robust procedures, and of sparse data description.

Research Interests
My theoretical research interests have focused on the mathematics of statistical inference and on theoretical questions arising in applying harmonic analysis to various applied problems. My applied research interests have ranged from data visualization to various problems in scientific signal processing, image processing, and inverse problems.

Personal Website
stat.stanford.edu/~donoho/

Gray, a renowned genomics expert at Lawrence Berkeley National Laboratory who helps lead a consortium of the nation’s top cancer researchers through the Stand Up To Cancer initiative, will expand on the institute’s lead in making personalized cancer medicine a reality for all patients

Donoho is a mathematician who has made fundamental contributions to theoretical and computational statistics, as well as to signal processing and harmonic analysis. His algorithms have contributed significantly to our understanding of the maximum entropy principle, of the structure of robust procedures, and of sparse data description.

Research Interests
My theoretical research interests have focused on the mathematics of statistical inference and on theoretical questions arising in applying harmonic analysis to various applied problems. My applied research interests have ranged from data visualization to various problems in scientific signal processing, image processing, and inverse problems.