Gears are ready-to-use applications that automate routine tasks, including metadata extraction, classification, quality assurance, format conversion, and full analytic pipelines. Because Gears are containerized algorithms that process data within the Flywheel platform, you can easily control dependencies and support reproducibility.
This article gives an overview of the types of Gears available in Flywheel, where to find Gears, and how to run them on your data.
Utility Gears are either a conversion Gear, which takes data from one format to another, or a QA Gear, which provides quality assurance on the data. Utility Gears can also be called pre-processing Gears and can be assembled into a pipeline to run automatically.
Here are a few examples of common utility Gears:
DICOM-MR-Classifier: Extracts DICOM header metadata and determines measurement classification. Works with Siemens, Philips, and GE DICOM data.
DICOM to NIfTI conversion (with PyDeface): Implementation of Chris Rorden's dcm2niix tool for converting DICOM (or PAR/REC) to NIfTI (or NRRD), with an optional implementation of Poldrack Lab's PyDeface to remove facial structures from NIfTI.
BIDS Pre-Curation: Prepare project for BIDS Curation. BIDS Pre-Curate offers a simple way to modify labels and classifications of project data to be compatible with the BIDS-spec.
Analysis Gears perform advanced preprocessing and data analysis. The outputs from the Gear is saved as a discrete analysis object, which you can view in the Analyses tab of the project.
Example Analysis Gears:
FreeSurfer 7.1.1: run recon-all: This gear takes an anatomical NIfTI file and performs all of the FreeSurfer cortical reconstruction process. Outputs are provided in a zip file and include the entire output directory tree from Recon-All.
VPNL: fLoc - Face Localizer Analysis Pipeline: Automated analysis of fMRI data from fLoc funcional localizer experiment used to define category-selective cortical regions.
All published gears are listed in the Flywheel Gear Exchange.
Not all Gears listed in the Gear Exchange are installed on your Flywheel site. To see what Gears are available to you, click Installed Gears in the left navigation menu.
Your Flywheel Site Administrator can help facilitate the installation of a new Gear.
Once you have installed a Gear on Flywheel, there are two ways to run Gears on your data: Manually or automatically.
If you have just a few acquisitions you want to process with a Gear, you can select the sessions and run a Gear from your project. The example below shows how to manually batch run a gear on multiple sessions. Learn more about how to run a gear
Each project in Flywheel can be configured to use Gear Rules. Gear Rules allow for customized automation of a project's data processing. Whenever the conditions of the gear rule are met, the rule triggers a specific gear to run.
For example, the Gear Rule below runs the DICOM MR Classifier Gear whenever a DICOM file is uploaded to the project:
When any acquisition file is added or modified (including classification and metadata), all files in the acquisition will be compared to all Project Gear Rules to determine if any new gears should be run on the data in the acquisition. One common use case for Gear Rules is to automatically process new data as it is uploaded to the project.
Flywheel records derivative information when gears are run to support the consistency and reproducibility of your project, including version, inputs, status, and outputs. This provenance makes it easy to ensure that all necessary processing steps were performed and performed consistently. To view this information, select the Provenance tab in a project.
Explore our webinars and videos for more in-depth training.