The image processing group in the Department of Molecular Structural Biology of the Max Planck Institute of Biochemistry is offering B.Sc, M.Sc./Diploma, and Ph.D.~thesis opportunities for students to work on three-dimensional image reconstruction, parallel visualization, and pattern recognition problems arising in cryo-electron tomography.

You should be working towards a degree in computer science/bioinformatics, mathematics/statistics, physics, (bio-)chemistry, biology, or electrical or mechanical engineering, have a sound mathematical basis and good knowledge of a high-level programming language, preferably C/C++ or Java in UNIX/Linux environments. Experience in one or more of the following fields is an advantage: biological imaging, pattern matching and machine learning, image or signal processing, computer graphics and computer vision, or Monte Carlo simulations. Ph.D. fellowships are available for qualified candidates.

The Max Planck Institute of Biochemistry, located in Martinsried, adjacent to the Grosshadern/Martinsried Life Sciences campus of the University of Munich, is one of the world’s leading research institutions in the life sciences. Our group works closely with biologists, (bio-)chemists, and physicists both in the department and internationally and offers a truly interdisciplinary experience in computational biology. Working languages are German and English. For further information, please contact

Dr. Christoph Best, best [at] biochem.mpg.de, Tel. +49-89-8578 2634 MPI für Biochemie, Am Klopferspitz 18, 82152 Martinsried, Germany

Please add the keyword “IP2005″ to the subject line to avoid being caught in our spam filter. Further information on possible research projects is available below.

POSSIBLE THESIS PROJECTS

Our group is concerned with image processing, reconstruction, visualization, and analysis problems arising in three-dimensional cryo-electron tomography.

In cryo-electron tomography, transmission electron microscopy images are taken of biological samples frozen in amorphous ice. Using images taken at different angles, their 3D structure can be reconstructed numerically. Image resolution in the nanometers is attainable with these methods, especially when many images are combined, so subcellular structures and macromolecular complexes can be investigated.

Computational challenges in this area are the high noise level, which requires sophisticated denoising and pattern recognition methods, as well as the structural variability of the objects and the problem of combining many images into one reconstruction. They offer a unique opportunity to combine

We have a state-of-the-art visualization facility based around a 24-processor SGI Prism and a cluster of HP visualization workstations, as well as access to the national supercomputing infrastructure. The Max Planck Institutes, Germany’s basic research organization, offer a world-class research environment with ample opportunity for interdisciplinary work and international exchange. Thesis work can be arranged with the International Research School for the Life Sciences located at the lab as well as with the local universities.

Here are some of the projects in which theses are possible in our group:
* INTELLIGENT PATTERN RECOGNITION AND CLASSIFICATION
Using machine learning methods, we try to build systems that automatically recognize and classify projection images of macromolecular complexes and cellular subunits and then reconstruct the original three-dimensional shapes. This project requires a good understanding of machine learning theory and image processing.

* PARALLEL HIGH-PERFORMANCE VISUALIZATION FOR BIOLOGY
Implement, extend and help maintain image processing and visualization programs for parallel processing on the SGI Prism architecture with the aim to aid biological understanding of 3D structures using interactive visualization. Requires good programming and image processing skills.

* RECONSTRUCTION AND SEGMENTATION ALGORITHMS FOR TOMOGRAPHY
Investigate and implement fast parallel 3D algorithms for reconstruction, denoising and segmentation of tomographic images using pattern recognition and machine-learning techniques. Requires good knowledge of numerical mathematics and parallel algorithms.

Reference:
http://www.rzg.mpg.de/~cbest/positions.html