Using Artificial Intelligence to study ovarian development in model fish
Utilisation de l’Intelligence Artificielle pour étudier le développement ovarien chez les poissons modèles
Résumé
The study of organ development most often requires a step of quantification of the biological structures of interest, including measurements of size and number. The ovaries are made up of anatomical structures (follicles) within which female gametes (oocytes) are formed. In fish, these structures, whose diameter varies from 20 μm to more than 1000 μm, have the particularity of being constantly renewed during adulthood. For a long time, biologists have been working to decipher the growth dynamics of these structures but, despite recent progress in three-dimensional (3D) imaging, access to the total cellular contents of the ovaries remained limited due to the absence of efficient methods for analyzing microscopy images, particularly 3D images. The emergence, a few years ago, of Artificial Intelligence for image analysis made it possible to resolve this technological obstacle. We used open-source Deep Learning algorithms accessible to biologists, which simplifies the segmentation step and overcomes the methodological biases of classic 2D stereological approaches. These approaches provided us with quantitative data of unprecedented precision on the different oocyte populations in the Medaka, at different stages of development in normal and disturbed conditions. All of these data now allow us to better understand the dynamics of asynchronous growth of oocytes and to decipher the role of molecular players in determining fecundity. Nowadays, the explosion in the number of AI tools made available for the analysis of 2D and 3D images gives hope for new perspectives for the study of ovarian development in fish.
Domaines
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |