The LIRMM and its ICAR team (Image & Interaction) team contributed to a recent paper published in the journal Nature Scientific Data. The paper presents the Light My Cells database, an open-access resource of microscopy images designed to support the development of machine learning models capable of predicting fluorescence from unlabeled transmitted light images.
This initiative is part of the France-BioImaging national infrastructure, which brings together more than 20 imaging platforms in France. By organizing bioimaging challenges, the infrastructure aims to build community, produce open databases, and foster the development of new image analysis methods.
In this context, the Light My Cells database comprises 2,574 acquisition series totaling 56,984 2D microscopic images. It covers a wide variety of biological samples and imaging modalities, bringing together the results of 30 studies conducted at 8 centers within the infrastructure.
Each acquisition series includes a transmitted-light image paired with at least one fluorescence image labeling key subcellular structures such as the nucleus, mitochondria, tubulin, or actin.
The collection is accessible via BioImage Archive (an EMBL-EBI data resource) and supports a wide range of applications, such as in silico labeling, segmentation, and cell profiling from unlabeled images. The data are provided in a standardized format (OME-TIFF) and accompanied by rich metadata compliant with international REMBI recommendations.
This work was conducted at the LIRMM by Dorian Kauffmann, under the supervision of Emmanuel Faure and Guillaume Gay, in collaboration with 28 co-authors nationwide.
Contact : Emmanuel Faure
LIRMM – département INFO
Équipe ICAR
Équipe France BioImaging Challenges
Université Montpellier
CNRS
