Preclinical imaging workflows have been growing in complexity resulting in generation of large datasets. We provide on the development of an open-source preclinical imaging XNAT-enabled informatics (PIXI) platform to manage the workflows of preclinical image data acquisition, capture imaging-associated experiments including metadata and annotations, and to implement computational pipelines in a unified environment. Our vision for PIXI extends beyond the initial implementation to support a federated network of PIXI instances and a PIXI Center to enable data sharing and collaboration across institutions.

PIXI is based on the widely used Extensible Neuroimaging Archive Toolkit (XNAT) as the underlying informatics architecture. The PIXI platform includes: the PIXI Server, which provides core database, user management, visualization, and workflow functionality; PIXI Point-of-Service (PoS) interface for data entry and PIXI Notebooks for data exploration and analysis; and PIXI Apps to enable automated image processing pipelines through Docker container environment. With the recent release of PIXI as of February 2024, preclinical positron emission tomography (PET), computed tomography (CT) and magnetic resonance (MR) DICOM images are pushed to the PIXI server for workflow management with metadata captured through the PIXI PoS and DICOM image files for search and reporting. Multi-mouse images are supported by a custom RESTful API and Docker pipeline facilitating the splitting of images into individual datasets. Imaging workflow information can be entered or edited through the PoS interface including animal modeling information/metadata, descriptive data, and drug therapies. In addition, we developed workflows to support upload and management of native Inveon PET and CT images and IVIS bioluminescence (BLI) images. XNAT’s search and reporting capabilities have been extended to support PIXI’s new data types and metadata. Importantly, we collaborated with the Open Health Imaging Foundation (OHIF) to expand the OHIF viewer’s capabilities to support 4-dimensional image visualization and analysis of preclinical images. Finally, we have successfully integrated Jupyter notebooks into the PIXI platform by using JupyterHub, providing researchers with seamless access to a secure, cloud based Jupyter notebook environment within PIXI. Additionally, retrieving PIXI database metadata and information from within a Jupyter notebook is made easy with XNATpy, an open-source XNAT client for Python. The integration with JupyterHub leverages Docker containers and Docker Swarm for efficient management and scalability of computing resources for PIXI Jupyter notebook users.  Building upon the Jupyter notebook integration, we have integrated Python dashboards directly into XNAT. Leveraging technologies such as Dash, Panel, Streamlit, and Voilà, this integration provides researchers and other technical users with tools to create and share interactive visualizations and applications that can then be started by general users from within the XNAT UI. This eliminates the need for users to interact directly with Python code.

Overall, the development of the PIXI platform is expected to have a profound impact on the management of preclinical imaging datasets and co-clinical imaging to support cloud-based computational pipelines and integration with multi-scale correlative biology to support precision oncology. Since PIXI was released in February 2024, PIXI has over 60 users at various levels of activity, and we hope to grow the user base as we continue to disseminate and expand the capabilities of PIXI. Additional information, including the free download of PIXI, instructional videos, documentation, and mailing list sign-up, is available at https://www.PIXI.org/.

Image/Figure:

Click to view full size

Image/Figure Caption:

Overview of the PIXI platform.  1) The PIXI server, 2) the PIXI Point of Service (PoS) environment, 3) PIXI App environment. 

Author