SciJava Ops: an improved algorithms framework for Fiji and beyond

• DOI: 10.3389/fbinf.2024.1435733
• Publication Date: 2024-09-27T05:10:35Z
• Authors: Gabriel J. Selzer; Curtis T. Rueden; Mark C. Hiner; Edward L. Evans; David Kolb; Marcel Wiedenmann; Christian Birkhold; Tim-Oliver Buchholz; Stefan Helfrich; Brian Northan; Alison Walter; Johannes Schindelin; Tobias Pietzsch; Stephan Saalfeld; Michael R. Berthold; Kevin W. Eliceiri
• Keywords: Implementation; Plug-in; Benchmarking; Interface (matter)

Decades of iteration on scientific imaging hardware and software has yielded an explosion in not only the size, complexity, heterogeneity image datasets but also tooling used to analyze this data. This wealth analysis tools, spanning different programming languages, frameworks, data structures, is itself a problem for analysts who must adapt new technologies integrate established routines solve increasingly complex problems. While many “bridge” layers exist unify pairs popular there exists need general solution existing toolkits. The SciJava Ops library presented here addresses through two novel principles. Algorithm implementations are declared as plugins called Ops, providing uniform interface regardless toolkit they came from. Users express their needs declaratively Op environment, which can then find available demand. By using these principles instead direct function calls, users write streamlined workflows while avoiding translation boilerplate bridge layers. Developers easily extend introduce libraries more efficient, specialized algorithm implementations, even immediately benefitting workflows. We provide several use cases showing both user developer benefits, well benchmarking quantify negligible impact overall performance. have initially deployed Fiji platform, however it would be suitable integration with additional platforms future.