Preclinical Imaging XNAT-enabled Informatics (PIXI) Center

ABSTRACT Preclinical imaging is widely used in cancer research to devise novel tumor detection strategies, assess tumor burden and physiology/biology, as well as to validate novel therapeutic strategies and predictive biomarkers of response to therapy. The typical imaging workflow is complex and may include multi-modality, multi-parametric imaging, multiple animals, and linked biology experiments. When multiplied by the number of animals per group in an experiment, multitudes of interventions (e.g., drugs), and the number of time-points in a longitudinal imaging protocol, the resulting datasets are vast and prohibitive to track and manage long-term. Non-trackable data results in poor reproducibility and presents obstacles for data mining and open science collaboration. To address this unmet need, we developed the Preclinical Imaging XNAT-enabled informatics (PIXI) platform. PIXI (v. 1.0.0) (available at https://www.PIXI.org/) was released for public use in February of 2024. PIXI v1.0.0 captures preclinical DICOM images and associated metadata. DICOM is not universally used nor adopted by preclinical imaging vendors as many vendors use proprietary image file formats, resulting in fragmented adoption. In Aim 1, we will extend the PIXI Server to capture non-DICOM preclinical imaging experiments, associated metadata, image-derived measurements, and linked-biology experiments to advance NCI’s precision medicine initiative (PMI). In PIXI v1.0.0, we additionally developed PIXI’s containerized application environment, integrated Jupyter notebooks to enable deployment of computational pipelines, and integrated PIXI Dashboards. As preclinical imaging data is becoming more complex with big data needs, the need for reproducible and unified analysis workflows is critical. In Aim 2, we will develop PIXI Dashboard services and computational imaging pipelines to enable visualization, processing, and analysis of preclinical imaging biomarkers and linked biology experiments, supporting NCI’s PMI objective of consistent processing and analysis of imaging biomarkers. Finally, the increasing complexity, volume, metadata needs, and big data needs of oncologic preclinical imaging workflows necessitate a domain specific preclinical imaging repository to support NIH's Policy for Data Management and Sharing (DMS). The NIH provides several domain-specific repositories for DMS, but none are dedicated for preclinical imaging nor with robust support for metadata and quality assurance needs. PIXI is uniquely positioned as a preclinical imaging DMS. In Aim 3, we will establish PIXI Center: A domain-specific data management and sharing (DMS) resource for oncologic preclinical imaging, enabling Centralized Learning in oncologic translational imaging research. The fourth aim of the proposal focuses on resource dissemination and training strategies. Overall, the renewal proposal will support the complex and growing demands in preclinical cancer imaging. We will enhance and extend PIXI’s capabilities to support additional experiments, extend metadata, develop and enable computational pipelines, and establish a much-needed domain-specific DMS, ensuring continued support, development, and wider adoption of PIXI to ultimately support NCI’s missions.