The team has close links with the Turing Center Multi-engineering platform where 3 engineers are specialized in image processing, database management and software development. The team develops inference
methods and machine learning tools for developmental biology, with the aim of understanding the relationships between morphogenesis and cell differentiation trajectories by bridging microscopy and single cell-omics.

Three expertise of the Team
  • Image processing
  • Machine learning
  • Single cell omics

The SIMS Group (Signal, Image and Sound) develops its research in the field of statistical signal and image processing. Tackling ill-posed inverse problems such as deconvolution and diffraction tomography is one of our main fields of expertise. We also focus on computational imaging issues, relying on mathematical tools from both optimization and Bayesian simulation fields. The optimal co-design of computational imaging systems is among our growing topics of interest, with the viewpoint of information theory. In the area of fluorescence microscopy, we investigate the superresolution capacity of random illumination microscopy, in collaboration with Institut Fresnel (Marseille) and the Centre for Integrative Biology (Toulouse).

Expertise of the Team:

  • Inverse problems
  • Computational imaging
  • Machine learning
The CeDRE team, led by Jacques Pécréaux (CRCN CNRS) is multidisciplinary, teaming up biologists, physicists and mathematicians. They are interested in mechanical aspects of cell division to understand its dynamical aspects and ultimately the extra-ordinary faithfulness of mitosis. Therefore, the team finely quantifies biological phenomena to recapitulate the results using physical models. In particular, the team performs advanced data analysis and modelling. It includes the fine tracking of spots like the tips of the microtubule, or super-resolution positioning of the centrosomes at a high frame rate (tens of frames per sec.). The obtained tracks are analysed statistically (for example, to classify the dynamics, reduce the dimension using PCA) or using Fourier spectral analysis to fingerprint the underlying mechanisms. To take full advantage of this data analysis, we model the sub-cellular mechanics during mitosis, in particular, the one grounded on microtubules, molecular motors and associated regulators on the one hand. On the other hand, we use agent-based simulations, especially cytosim, to screen for the variety of behaviours without solving the analytical equations. Finally, the team contributes to developing autonomous microscope by enslaving the driving of the microscope to on-the-fly analysis of acquired images, in particular using machine and deep learning.

About

The Image Analysis Hub is an open access, equal access core facility committed to offering support in image analysis. Our webpage is: https://research.pasteur.fr/en/team/image-analysis-hub/

What we do.

As part of the C2RT, we strive to ensure the continuity between image acquisition and image analysis. To this end we rely on our expertise in imaging and collaborate with other facilities such as the UTechS-PBI and UTechS-UBI when pertinent. All requests involving images are considered.

Our services follow four axes:

1. Offer walk-in support and trainings for questions involving image analysis.

This activity aims at offering to users quick answers to scientific questions involving well-established pipelines, for which a commercial or published tool exists and can be used conveniently. Users can address their question to the facility during open-desk sessions or directly via one-to-one requests. Depending on the effort involved, the solution is derived and proposed onsite, or individual  trainings are scheduled. For general topics, the Hub organises regular courses and workshops, possibly involving external teachers or providers.

For instance, see below for the announcement of our open-desk, organised regularly every two week.

2. Build and deploy custom analysis tools for projects requiring special developments.

Research endeavours to address original questions, for which analysis tools might be lacking or incomplete. The Image Analysis Hub aims at creating or implementing novel tools based on existing algorithms to address these questions, using skills in image analysis and software development. More than just developing the analysis tool, this activity often involves deriving a suitable analysis methodology, for which the facility expertise in microscopy and biophysics is key. Engineers work in close collaboration with users within the framework of a scientific project over medium or long durations. For projects whose effort would extend beyond typical facility usage or involve original research work, the project may be directed to the BioImage Analysis unit after a discussion with all parts.

3. Maintain an infrastructure for autonomous image analysis. Deal with complex tool deployments.

Data volume and modern analysis techniques may call for a computing power not always present in Pasteur labs. Providing open-access workstations unlock barriers to compute-intensive tools. They also act as the central sharing points for commercial softwares, making them available to the whole campus. Finally, some specialized tools require special deployment efforts, e.g. to make such a tool able to exploit the HPC infrastructure of the Institut Pasteur.

4. Develop original and innovative software tools for image analysis, whose scope exceed user projects.

Software development and image analysis skills of the facility can be leveraged to build ambitious software tools shipping innovative technologies. These tools exceed the scope of single projects and address the unarticulated needs of the Pasteur community and their creation is part of the development activity of the facility.

The scientific project of the BioImage Analysis (BIA) unit is to develop image analysis and computer vision tools for the processing and quantification of multi-channel temporal 3D sequences in biological microscopy. The topics are centered about the development of new algorithms for multi-particle tracking, deformable models, mathematical imaging and spatial distribution analysis. The group has produced powerful tools for spot detection and counting in real-time imaging of virus and genes, movement and shape analysis in 3D+t microscopy and cell growth analysis. These methods and algorithms have now been regrouped under the open-source and free platform Icy (http://icy.bioimageanalysis.org), which provides a comprehensive framework for extended reproducible research in bioimage informatics. They have been instrumental for the successful achievement of a large number of collaborative biological projects.