Implementation of LSIDs in the Open Microscopy Environment

Biological microscopy has always required an 'imaging' capability: traditionally, an image of a sample was drawn on paper, or with the advent of light-sensitive film, recorded on media that conveniently allowed reproduction. However, the application of digital detectors to microscopy has converted the biological microscope into a quantitative assay device. The clinical and research applications for digital imaging microscopy are enormous, but these applications generate large amounts of data/a single time-lapse image can easily be 500 Mbytes. Four years ago, we began developing a data management tool for digital biological microscopy. Our project, the Open Microscopy Environment (OME; http://openmicroscopy.org) is now the leading provider of open source image management software for biological microscopy. A critical capability in image data management is understanding the source and history of all data. Knowledge of how data was recorded and the specification of the image acquisition system is especially important in biological imaging, as there are a large range of different imaging modalities. We have defined a common data model for biological microscopy that provides the framework for storing and sharing this image data. This structure is referred to as the OME Data Model, and is being used by a number of commercial and independent entities as a biological image data standard. However, the OME Data Model can capture a large amount of information, and it is critical that as much of this information be filled out as possible. Unfortunately filling out complex data forms is time consuming and thus unlikely to be adopted by most users unless it is automated and simplified. A separate project has specified a system of Life Science Identifiers (LSIDs) as unique identifiers for biological data and systems. This proposal seeks funds to support the integration of LSIDs into the OME specification for microscopes. Most importantly, we will build a user tool that stores microscopes configurations and then obtains an LSID for a microscope system that identifiees that system and can be used to describe most of the parameters for a given image acquisition system. A complex instrument specification and configuration can then be indicated by a single mmouse-click. Most importantly, this specification is uniquely identified, stored with all recorded images, and accessible ('resolvable'') by any OME server or other LSID-compliant system.

Biological imaging is now used as an assay technique, most recently in time-lapse multichannel fluorescence imaging of cellular dynamics and in high-content genomic or small molecule screens. Maintaining critical contextual descriptions (e.g., cell lines, mutants, microscope configuration), the relationships between images, annotations and quantitative analyses while preventing data loss during file format migration or data exchange with collaborators is a major challenge for any cell biologist. Our project, the Open Microscopy Environment (OME; http://openmicroscopy.org) solves this problem with a Data Model that simultaneously defines the relationships between complex data types and provides a mechanism for changes to the Data Model to support changing experimental requirements. The OME Data Model contains explicit support for unique identifiers for microscopes, detectors, filter sets, and objective lenses (http://www.openmicroscopy.org/XMLschemas/OME/latest/ome_xsd/), however it is impractical to expect users to remember these IDs or even enter them as additional data into every image file. Currently microscope software manufacturers only support serial numbers for their microscope components which are not 'resolvable', or convertible to tables of appropriate system description and configuration. This project will develop a tool that will allow users to 'register' their microscope systems, obtain LSIDs, and then use these LSIDs to fill out image metadata fields by simply indicating the image acquisition they used. The result will be improved specification of image metadata, increasing implementation LSIDs, elaboration of the OME Data Model, and improved data provenance and thus better tracking of the sources of biological image data. Moreover, the source of any image can easily be accessed and examined using an LSID resolver, either within an OME server instance, or other LSID-compliant software. This is a short, targeted one year project that will supply a small but critical tool to the imaging community and leverage the existing Data Model infrastructure in OME