How do you turn 25 years of medical imaging into something researchers can actually use? At Bio-IT World, Starfish Chief Science Officer Ari Burman shared the answer, built with Arizona State University.
ASU’s Tsimane Health and Life History Project studies healthy aging in an Indigenous Bolivian community with some of the lowest rates of coronary artery disease ever recorded. But two decades of CT scans lived on hard drives that were shipped by mail and curated by hand. Each researcher request triggered weeks or months of waiting, complicated by inconsistent DICOM metadata and strict consent rules that ruled out the cloud.
The team rebuilt the workflow on Starfish. An automated pipeline replaced the hard-drive handoff. A Python process normalized 25 years of messy metadata, auto-detected organ types, and made every scan queryable by body part, quality, and reconstruction type. Zero-copy linked views kept the data in place, on-premises, and fully governed.
The payoff: months of manual fulfillment became minutes of self-service search, with complete audit trails intact. The scripts are open-sourced on GitHub, and the project earned a Bio-IT World award.
Read the full white paper here.
