BioImageIT: Open-source framework for integration of image data management with analysis

Sylvain Prigent, Cesar Augusto Valades‐Cruz, Ludovic Leconte, Léo Maury, Jean Salamero, Charles KervrannNature Methods 2022 · DOI · OpenAlex

Introduction

In the rapidly evolving field of bioimaging, researchers frequently encounter a significant "silo" problem. While powerful tools exist for image analysis (e.g., Fiji, Icy, CellProfiler) and robust systems for data storage (e.g., OMERO), these components often lack interoperability. This fragmentation forces scientists into labor-intensive "janitorial" tasks: manually transferring massive datasets, writing custom scripts to bridge software gaps, and struggling to maintain the metadata necessary for reproducible research. As microscopy becomes increasingly data-intensive, particularly with the integration of AI and deep learning, manual data management is no longer sustainable.

BioImageIT addresses this challenge by providing an open-source framework that acts as a "computational overlay," bridging the gap between data management and complex analysis workflows. It integrates existing tools into a single, FAIR-compliant (Findable, Accessible, Interoperable, and Reusable) ecosystem accessible via both a user-friendly graphical user interface (GUI) and a powerful Python API.

Framework Architecture and Design

BioImageIT is built on a modular, three-layer architecture designed to supplement rather than replace existing software. This structure allows it to scale and adapt to various laboratory needs.

BioImageIT architecture and LLSM workflow demonstration. (a) The three-layer architecture including client applications, the core API, and back-end plugins for data management and processing. (b-i) Application of the framework to a Lattice Light Sheet Microscopy (LLSM) workflow, showing data acquisition, processing steps (deskewing, deconvolution, tracking), and final 3D visualization.

As illustrated in Figure 3a, the framework consists of:

  1. Back-end Plugins: These connect the system to diverse data sources (local files, remote servers, OMERO) and processing "runners" (local environments, Docker containers, or remote clusters).
  2. Generic API Layer (Python): This provides a programmatic interface for data scientists to automate pipelines and build complex workflows.
  3. Client Applications: A GUI and Jupyter-based interface cater to biologists who prefer assisted, step-by-step workflows.

A key innovation is the "wrapper" strategy. By using simple XML/JSON configuration files to define tool interfaces, BioImageIT allows algorithms written in any language to be integrated without modifying the core framework. This approach lowers the barrier for developers to contribute new tools, such as deep learning models, to the community.

Key Insights: Reconciling Management and Analysis

BioImageIT treats data management and analysis as integrated processes. A critical feature is the automatic generation of metadata during analysis runs. This ensures that the "provenance"—the complete history of operations performed on a dataset—is preserved, directly supporting research reproducibility. The framework's persona-based design ensures it serves the entire lab: biologists use the GUI, data analysts utilize the Python API, and developers benefit from a standardized distribution method.

Results and Application

The authors demonstrated the framework's utility through a complex Lattice Light Sheet Microscopy (LLSM) workflow (Figure 3b-i). The pipeline successfully orchestrated multiple tools—including pycudadecon for deskewing and deconvolution, Trackmate for vesicle tracking, and napari for 3D visualization—to process high-dimensional data from live genome-edited HeLa cells. This demonstration highlighted BioImageIT's ability to handle data that is traditionally difficult to manage manually.

Beyond a prototype, BioImageIT is currently being deployed across ten imaging platforms within the France-BioImaging national infrastructure. It has proven capable of seamless interaction between different programming environments, such as running Python-based deep learning tools alongside Java-based tracking software.

Limitations and Future Directions

While robust, BioImageIT currently has limitations. It is not yet designed to handle patient-sensitive data, which requires more stringent security protocols. Future development aims to expand the Generic API to better support Java and REST APIs and to allow users to initiate analysis pipelines directly from within their primary database GUIs. The team continues to work on generalizing the tool for broader use with systems beyond the current OMERO integration.

Summary updated

BibTeX
@inproceedings{prigent2022bioimageit,
  author = {Sylvain Prigent and Cesar Augusto Valades‐Cruz and Ludovic Leconte and Léo Maury and Jean Salamero and Charles Kervrann},
  title = {{BioImageIT: Open-source framework for integration of image data management with analysis}},
  booktitle = {Nature Methods},
  year = {2022},
  doi = {10.1038/s41592-022-01642-9},
}
Figure