Purpose: Imaging tools for preclinical and clinical research are invaluable for detecting and monitoring disease and assessing treatment response non-invasively. The continuous generation of biomedical images and associated metadata at an increased rate makes image data management and sharing challenging. Especially for pre-clinical images stored in vendor-specific proprietary file formats, the routine image data handling and analysis methods become very cumbersome. Despite the availability of several image management tools, which are mostly tailored for clinical DICOM-based images, there is still a need for robust data handling tools. A web-based platform, BiRAT (Biomedical Imaging Registry, Analysis and Translation) is being developed to facilitate the pre-clinical image data registry and processing through early data capture, storage, processing, and analysis.

Methods: The application uses a data-driven bottom-up approach to capture and store images and associated experimental data in native formats. The original unstructured image data are annotated and uploaded through web and desktop clients in a structured hierarchical database to a cluster-based distributed data storage system. Various frontend and backend web applications are being developed to manage, access, and share data for better security and accessibility. The application also includes a robust image viewer. Future development will include modular processing software tools for computationally intensive tasks and AI implementation for various image data processing, such as auto-splitting multi-animal image acquisition, bed removal, image segmentation, analysis, and quantification applications.

Results: The framework for containerized server architecture is ­­­deployed incorporating advanced technologies to provide distributed and scalable data storage. The server is configured to manage containerized workloads through the latest Kubernetes distribution system. A web application suite is also being implemented, replacing the initial prototype development. The new implementation incorporates JavaScript programming for implementing robust, dynamic, data-driven web applications. This implementation also enables modularization, making developing specific new applications as needed and incorporating various third-party web applications easier. The current development focuses on developing a robust web-based database capable of performing fast data upload, comprehensive and easy data annotation, data retrieval, data sharing, and basic data process applications.

Conclusion: Robust web applications and cluster-based scalable servers are being developed, transforming traditional approaches into a centralized data storage management system that allows more advanced data archiving and processing pipelines.

Acknowledgment: This work is supported by grants U24CA264298 (Co-clinical research for imaging tumor-associated macrophages) and 5P30CA124435-10 (Stanford Cancer Institute Support Grant).

Image/Figure:

Click to view full size

Image/Figure Caption:

Figure 1: BiRAT web-based platform for preclinical image registry and processing design to accommodate the basic data workflow in preclinical imaging (Top Left). New web development is progressing, consisting of fast data upload, data drive data annotation, easy data browsing, filtering and retrieval, and data sharing capabilities (Top Right). Image-based viewing, processing, and analysis are also being implemented (Bottom Left and Right).  

Author