NiQuery: Stop searching your neuroimaging datasets. Start testing your hypotheses

The field of neuroimaging has witnessed an exponential increase in data production and availability, fueled by technical advances that have enabled large-scale data collection, as well as by changes in policies and funding initiatives that support open data-sharing programs. As the number of openly available datasets grows larger, the need for efficient, standardized, and flexible tools to index, inspect, and fetch neuroimaging data is becoming a pressing need. OpenNeuro is a popular data hosting service in neuroimaging research; it allows to share publicly neuroimaging data stored following the Brain Imaging Data Structure (BIDS) standard. Although tools such as bids2table (https://childmindresearch.github.io/bids2table/bids2table.html) or Neurobagel Query enable interrogating BIDS datasets (including those on OpenNeuro) and the associated metadata, researchers that require using a subset of one or multiple datasets stored in data repositories typically need to fetch the entire dataset, and to use dedicated tools to interrogate and select the relevant data, hindering interoperability and scalability. Furthermore, existing tools to interrogate BIDS datasets only enable interrogating the metadata, and miss other features that can only be obtained by inspecting the actual volumetric data. Conversely, command-line tools allowing to fetch datasets from remote servers (e.g., Cohort Creator) do not provide data inspection or metadata-based filtering capabilities. Thus, a tool to efficiently interrogate and filter neuroimaging data hosted on public repositories is missing.We present NiQuery , a principled, open-source Python library designed to address these challenges by providing a robust framework for querying neuroimaging datasets, inspecting metadata, performing subset selection, and aggregating results. NiQuery seeks to facilitate reproducible research by enabling transparent access to the content of neuroimaging datasets, including volume-specific metadata.