Deep learning for bioimage analysis in developmental biology

Abstract

Deep learning has transformed the way large and complex image datasets can be processed, reshaping what is possible in bioimage analysis. As the complexity and size of bioimage data continues to grow, this new analysis paradigm is becoming increasingly ubiquitous. In this Review, we begin by introducing the concepts needed for beginners to understand deep learning. We then review how deep learning has impacted bioimage analysis and explore the open-source resources available to integrate it into a research project. Finally, we discuss the future of deep learning applied to cell and developmental biology. We analyze how state-of-the-art methodologies have the potential to transform our understanding of biological systems through new image-based analysis and modelling that integrate multimodal inputs in space and time.

Tissue clearing and 3D imaging in developmental biology

“…They have two advantages: they can solve highly complex segmentation tasks and they scale well to the considerable size of the data. However, they have two weaknesses: their training is extremely slow and labor-intensive and they rarely work well outside of a narrow definition of the type of data they were trained for (reviewed by Hallou et al, 2021 ). Often, specific tasks of image analysis require custom pipelines, which are now regularly published ( Jin et al, 2019 preprint; Renier et al, 2016 ; Susaki et al, 2020 ; Wang et al, 2021 ; Young et al, 2020 ), but require the use of scripts and user interventions on the source code, putting them out of reach of most users.…”

Towards Tabula Gallus

“…Combined with whole-mount immunostaining or in situ hybridization, still-improving tissue clearing methods to render organs and tissues transparent can provide molecular profilings at a cellular or subcellular resolution and help to create 3D and 4D body expression maps [ 129 ]. In addition, applying artificial intelligence and machine learning algorithms to address image analytical issues will also be used [ 130 ].…”

Advances in live imaging early mouse development: exploring the researcher's interdisciplinary toolkit

“…cell tracking in 3D for thousands of cells across hundreds of time points). Machine learning-based approaches to automate analysis are proving effective (see Hallou et al, 2021, in this issue), and several software packages have been developed for such image processing and 3D cell-tracking requirements, including MaMut (Wolff et al, 2018), BigDataViewer (Pietzsch et al, 2015) and RACE (Stegmaier et al, 2016).…”