Menu
Deep Learning for Cell Tracking on 3D + time live embryos

Deep Learning for Cell Tracking on 3D + time live embryos

By In Aiv Internship On July 12, 2018


Internship title: Deep Learning for Cell Tracking on 3D + time live embryos

LABORATORY
Name: LIRMM
Affiliation: CNRS
Address: Montpellier
E-mail: emmanuel.faure@irit.fr

LAB Director
Name: Phllippe Poignet
Phone number: 0467418509
E-mail: direction@lirmm.fr

SUPERVISOR
Name: Emmanuel FAURE
Phone number: 0652972406
E-mail: emmanuel.faure@irit.fr

Subject Keywords: Deep Learning, Data Scientist, Image Processing, Developmental Biology
Tools and methodologies: Deep Learning,Image Processing
Summary of lab’s interests: The intern will be based in Montpellier, and the supervision will be provided by a computer scientist Emmanuel Faure (CNRS) and a biologist Patrick Lemaire (CNRS)
Mails : emmanuel.faure@irit.fr and patrick.lemaire@crbm.cnrs.fr

Emmanuel Faure, (LIRMM Montpellier from Fall 2018 https://www.lirmm.fr/icar/) is a Data Scientist using recent machine learning techniques to understand quantitative biology and medicine.

Patrick Lemaire’s team (CRBM, Montpellier, France; http://www.crbm.cnrs.fr/en/recherche/equipes/)focuses on animal embryonic development, using ascidians as a model system.
Project summary: Machine Learning and more specifically Deep Learning give very promising results in various domains of biological research. The aim of the internship is to apply Deep learning to the field of biological imaging, and more specifically to 4D (3D plus time) cell segmentation, the recognition and tracking of each cell and its progeny in live developing embryos. The internship will start from a preexisting database of more than 10 fully-segmented embryos (>10000 tracked cells). We would like to create an end-to-end method for complex 3D cell tracking using deep learning. In this internship, depending on the profile and interests of the intern, we can:
– Implement and adapt to our dataset existing end-to-end deep learning methods for instance segmentation (the identification of individual objects).
– Explore theoretical aspects of the adaptation of the memory concept in Recurrent Neural Networks (LSTM) to cell tracking in 3D+time images.
– Use deep learning to predict and define the rules of cell division during embryogenesis.
This 5-6 months internship can lead to a PhD project, potentially combining computer approaches with the experimental validation of predictions.

Interdisciplinary aspect of the project: The internship is mainly intended for students with an initial training in maths and computer science, though biologists with very good skills in computer programming and statistical analysis could be considered. An interest in biology is welcome. Team spirit, autonomy, dynamism, creativity would be appreciated. Proficiency in technical English or French must be sufficient to write technical documentation, and to interact verbally daily.
Required programming language: python
The selected student should basic knowledge of at least one of the following topics: machine learning theory, image processing, high performance computing or the use deep learning libraries (e.g. Keras).