gms | German Medical Science

Introduction: Integrating picture archiving and communication systems (PACS) into clinical environments marks a significant advancement in the management of medical imaging data [1]. However, the emerging interest in machine learning and its application across diverse settings have sparked the demand for PACS that support research purposes beyond conventional use. Addressing this demand, we introduce PACS2go, a portable research PACS providing the necessary infrastructure and interfaces for storing and exchanging imaging data in a research context. PACS2go is specifically designed to support multiple medical imaging formats and annotate data with metadata for enhanced research utility.

Methods: PACS2go's architecture integrates a file storage system using the open-source imaging informatics platform XNAT [2], a metadata store using PostgreSQL [3], and a user interface developed with the Python web framework Plotly Dash [4]. Key design principles include portability, simplicity, and an emphasis on enhancing data utility for interdisciplinary medical research collaboration. Additionally, the introduction of a hierarchal data model simplifies metadata management. All modules are containerized using the open-source Docker platform and are orchestrated using the docker-compose tool, ensuring ease of setup and consistency across different environments.

Results: The system supports diverse medical imaging formats, including DICOM, NIfTI, and TIFF, offers robust metadata management features, and facilitates collaborative research through its intuitive project management interface. The practical application of PACS2go was demonstrated using the 2019 Kidney Tumor Segmentation Challenge (KITS19) dataset, a roughly 19 GB large collection of CT imaging and segmentation masks for 300 patients who underwent nephrectomy for kidney tumors [5]. The KITS19 dataset has been successfully uploaded to, downloaded from, annotated, and viewed using PACS2go, emphasizing the platform's capability to handle larger-scale data in real-world research settings.

Discussion: PACS2go addresses the challenges associated with managing and exchanging medical imaging data in a research context. PACS2go's all-in-one application design considerably simplifies the user experience, supporting various medical imaging data formats for storage, organization, and viewing. Additionally, PACS2go's open-source nature allows for the extension of metadata capabilities to meet specific research needs.

Nonetheless, PACS2go is still in its infancy, facing challenges that highlight areas for future development. Specifically, PACS2go's claim to simplicity is yet to be validated through case studies, and a more sophisticated approach to versioning and integration into machine learning pipelines is desirable.

Conclusion: Our open-source research PACS PACS2go stands out for its portable approach to support collaborative, interdisciplinary medical research and offers robust data management and metadata enrichment features for various file formats. By addressing its current limitations and building on its strengths, we aim to push forward to provide the necessary infrastructure and interfaces for the storage and exchange of data focusing on medical research applications. PACS2go is available on GitHub via the following link: https://github.com/frankkramer-lab/pacs2go.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.