Guo C1, Endrei M1, Narayanan A1
1University of Queensland Research Computing Centre, Brisbane, Australia
Biography:
Chenxi Guo is a Research Software Engineer at the University of Queensland's Research Computing Centre, where he focuses on XNAT plugin development for medical imaging infrastructure. His current work bridges institutional Hierarchical Storage Management and the XNAT platform, supporting researchers across the Australian Imaging Service ecosystem.
Chenxi joined UQ RCC in 2025. Prior to that, he held Research Software Engineer roles at the University of Technology Sydney and the University of New South Wales, and taught Computer Science courses at the University of Wollongong and the University of Sydney. He completed his Master of Software Engineering at the University of Sydney in 2022.
His recent first-author publication, "EpiGIS Pro: an AI-powered geospatial intelligence platform for integrated disease surveillance and predictive analytics" (International Journal of Health Geographics, 2026), reflects his broader interest in applying software engineering practices to interdisciplinary research problems — a perspective he brings to his work bridging storage infrastructure and imaging platforms.
This is his first presentation at eResearch Australasia.
Abstract:
XNAT is widely adopted for managing medical imaging data, but institutions running it at scale may use tiered storage — typically Hierarchical Storage Management (HSM) systems that automatically move older data from fast, higher-cost storage like SSDs to slower, cheaper archive storage like hard disks or tape. The XNAT interface gives researchers no indication of which scans sit in slower storage, producing unexpected delays of seconds to minutes. The problem generalises to any interactively accessed repository backed by tiered storage.
We developed an open-source XNAT plugin that queries the underlying filesystem to surface storage tiering information (Local, Recalling, Archived) in real-time for every scan on the session page. A complementary admin interface allows configurable per-resource thresholds, supporting four evaluation modes: worst-case, file-count, byte-size, and percentage. The plugin builds on standard XNAT plugin infrastructure, exposes a REST API for programmatic access, and integrates with the XNAT Container Service to trigger non-interactive or pipeline recalls.
The plugin has been evaluated in the UQ Research Computing Centre's XNAT environment, backed by the institutional HSM. Researchers can see at a glance which scans are accessible, and which require recall. Boundary tests across all four threshold modes confirm correct strict-greater-than semantics, including precision edge cases.
This work demonstrates a reusable architectural pattern for making tiered storage transparent to researchers without modifying the XNAT core. The same pattern extends to cloud deployments using block storage volumes with lower-cost object storage back-ends, offering data repository managers a path to more cost-effective use of their storage resources.
