In order to answer image data analysis demands, France-BioImaging is launching its first data machine learning competition: welcome to the Light My Cells challenge!

The challenge

The Light My Cells France-Bioimaging challenge aims to contribute to the development of new image-to-image ‘deep-label’ methods in the fields of biology and microscopy. Basically, the goal is to predict the best-focused output-images of several fluorescently labelled organelles from label-free transmitted light input-images. And we need you for that!

More information about the challenge: lightmycells.grand-challenge.org/lightmycells

How to participate

We have defined the challenge as a single task with two phases:

• A preliminary test phase (on 30 images) to familiarize with the algorithm submission procedure, with the possibility to have five submissions (with a maximum of one by week)
• The final test phase (on 300 images) with only one submission accessible will not give the possibility to evaluate their algorithms before submitting.

So, you have until the end of the first phase, on March 30, 2024, to register and participate at this Light My Cells challenge. Nonetheless, you can start working on your preliminary algorithm and tests on February 14st, 2024 (with the release of the training database)!

Register now! lightmycells.grand-challenge.org/lightmycells

Prizes

For top 3 winners:

• Award certificate
• A challenge paper will be written with the organizing team’s members for submission to journals
• Invitation to publish their methods in the proceedings of the IEEE International Symposium on Biomedical Imaging 2024s
• Support and integration of open source code into open science image processing and analysis software (e.g. BioImage Model Zoo, Napari)

For the 1st: 

• Invitation to 2024 France-Bioimaging annual meeting
• Graphic card
• Android tablet

For the 2nd:

• Graphic card
• Android tablet

For the 3rd:

• Android tablet

Why launching a challenge?

To develop powerful methods that will then end up in creating public databases, standards & benchmarks in the field of bioimaging! The FBI challenge is hinged on a double contribution: from core facilities engineers and from data scientists. The first group acquired a large number of images to build a dataset, that will later be used by the algorithms. These images were produced by microscopy engineers & technicians from FBI’s platforms. As for the second contribution, this is where the challenge starts! The challenge is then published to have a maximum of data scientists to work on the algorithms that best fulfill the analysis task.

The first project is also based on four pillars:

• Open source + FAIR (Findable, Accessible, Interoperable, Reusable)
• Supervised learning, it involves annotated datasets to maintain control over performances.
• In silico annotations, a computer labeling method to avoid manual annotation and its drawbacks.
• Image-to-image analysis tasks, an image analysis tasks which aim to predict an output image from the input one.

For any questions, please contact Dorian Kauffmann: dorian.kauffmann@france-bioimaging.org.