Providing open imaging data at scale: An EMBL-EBI perspective

Abstract

Biological imaging is one of the primary tools by which we understand living systems across scales from atoms to organisms. Rapid advances in imaging technology have increased both the spatial and temporal resolutions at which we examine those systems, as well as enabling visualisation of larger tissue volumes. These advances have huge potential but also generate ever increasing amounts of imaging data that must be stored and analysed. Public image repositories provide a critical scientific service through open data provision, supporting reproducibility of scientific results, access to reference imaging datasets and reuse of data for new scientific discovery and acceleration of image analysis methods development. The scale and scope of imaging data provides both challenges and opportunities for open sharing of image data. In this article, we provide a perspective influenced by decades of provision of open data resources for biological information, suggesting areas to focus on and a path towards global interoperability.

Keywords: Bioimaging; Data integration; Data management; Metadata.

Fig. 1

Growth in the total number of released datasets (top panel), and cumulative size of available image data (bottom panel) for EMPIAR (green) and the BioImage Archive (blue)

Fig. 2

Diversity in imaging applications/modalities. From left to right, top to bottom: focussed ion beam scanning electron microscopes (FIB-SEM; EMPIAR-10310), soft X-ray tomography (EMPIAR-10416), 2D electron microscopy (S-BIAD462), shortwave infrared imaging (S-BIAD548), confocal fluorescence microscopy (S-BIAD582), lattice light-sheet microscopy (S-BIAD573), whole-slide imaging for histology (S-BIAD419), widefield microscopy (S-BIAD531) and multichannel high-content screening (S-BIAD145)

Fig. 3

Describing an image in ways that make it useful for reproducibility and reuse often involves references and links to other types of biological information. Central image from S-BSST429, “The zebrafish as a novel model for the in vivo study of Toxoplasma gondii replication and interaction with macrophages” (Yoshida et al. 2020), with a corresponding image from EMPIAR-10460 at its right. Specialised data resources for these other biological domains can provide dedicated functionality, for example, BioSamples (Courtot et al. 2022) for sample management, the European Nucleotide Archive for genomic sequence data (Cummins et al. 2022) and the Flurescent Protein Database (FPBase) for fluorescent proteins (Lambert 2019)

Fig. 4

Display of an image together with metadata aligned with the REMBI model including image file structure, sample treatment and fluorescence channel content, from BioImage Archive accession S-BIAD144 (Hartwig and Höglinger 2021)