Fastr: A Workflow Engine for Advanced Data Flows in Medical Image Analysis

Abstract

With the increasing number of datasets encountered in imaging studies, the increasing complexity of processing workflows, and a growing awareness for data stewardship, there is a need for managed, automated workflows. In this paper we introduce Fastr, an automated workflow engine with support for advanced data flows. Fastr has built-in data provenance for recording processing trails and ensuring reproducible results. The extensible plugin-based design allows the system to interface with virtually any image archive and processing infrastructure. This workflow engine is designed to consolidate quantitative imaging biomarker pipelines in order to enable easy application to new data.