[OME2026] Workshop: ngio: Simple and Declarative OME-Zarr Processing in Python

Abstract: The emergence of OME-Zarr as a standard for large-scale microscopy data demands analysis tools that can exploit its architecture for efficient, scalable computation. We present ngio (Next-Generation I/O), an open-source Python library that provides a unified interface not only for reading and writing OME-Zarr files, but also for building end-to-end image processing pipelines directly on top of them.

ngio is built around three core feature areas. First, a simple object-based API allows users to easily open, explore, and manipulate OME-Zarr images and high-content screening plates, and to create or derive new images and labels with minimal boilerplate code. Second, rich tabular and region of interest support enables users to extract and analyze specific regions, and store measurements alongside image data. Third, scalable and declarative data processing is facilitated through powerful iterators for building generalizable image processing pipelines, complemented by an extensible mapping mechanism for custom parallelization strategies.

At the ngio workshop, you will get some hands-on experience with ngio to work with OME-Zarr images, interact with tabular data, and build declarative processing units.