Skip to content

Ingest a Structured Folder of Imaging Data using the CLI

The ingest folder command should be used if you are importing data that you previously downloaded from Flywheel, or if you are importing data that was shared from another Flywheel site.

The folder import command uses the folder structure to map your data to the Flywheel hierarchy. Learn how to run the ingest folder command.

Below is the basic Flywheel hierarchy:

Folder structured for import folder


fw ingest folder [optional arguments] [SRC]

Instruction Steps

Required Arguments

Required argument Description
SRC The parent folder for the directory you want to import

De-identify Data

If you are using  project, group, or site de-id profiles to de-identify data, verify that your DICOMs are not compressed in a zip files because the ingest folder command cannot de-identify zipped DICOMs.  Use the --dicom flag to indicate the folder with your images to be zipped before upload.

For example, let's say you have the following file structure:

└── Data

   └── SUB-001

        └── SES-001

            └── acq-001

                └── dcm001

                    ├── IM-0001-0001.dcm

                    ├── IM-0001-0002.dcm

The command for de-identifying those DICOMs using a de-id profile looks like:

fw ingest folder -g groupid -p Projectname --dicom dcm001 ~/Desktop/MyData

Optional Arguments


If you are using multiple optional arguments in your command, consider creating a config file. See our article for more information on creating and using a config file.

Optional arguments Description
-h, --help show this help message and exit
--compression-level The compression level to use for packfiles -1 by default. 0 for store. A higher compression level number means more compression
--copy-duplicates Upload duplicate files to a new project when using --detect-duplicates

The project created for the duplicates will be named [target project label] _TimestampOfIngest. The duplicates project will have the same permissions as the original destination project. No additional project artifacts need to be copied (gear rules, data views, description, etc).
--de-identify De-identify DICOM files, e-files and p-files prior to upload.  Applies custom de-identification when used with a deid-profile set in a configuration file.  By default, it will apply the minimal de-identification to DICOM files:
Remove Patient ID, Patient Name, and Patient Birthdate
Patient Age convert to months
--de-id profile NAME Use the De-identify profile by name. Use this flag if you have multiple de-id profiles in a single config.
--detect duplicates Flywheel scans both the source dataset and any data already in the Flywheel project. Flywheel looks for the following:
File path conflict in the source dataset- File path in the upload dataset is not unique
File path conflict in Flywheel - file already exists
Duplicate StudyInstanceUID in source dataset
Duplicate StudyInstanceUID in Flywheel - UID already exists
Duplicate SeriesInstanceUID in the source dataset
Duplicate SeriesInstanceUID in Flywheel - UID already exists
* Duplicate SOPInstanceUID in series
--detect-duplicates-project DETECT_DUPLICATES_PROJECT Specify one or multiple project paths to use for detecting duplicates
--dicom NAME The name of dicom subfolders to be zipped prior to upload (default: dicom)
--enable-project-files Enable file uploads to a project container
--encodings ENCODINGS Set character encoding aliases. E.g. win_1251=cp1251
--exclude PATTERN Patterns of filenames to exclude. For example:
Exclude a single file:
Exclude all files of a specific filetype
--exclude-dirs PATTERN Patterns of directories to exclude (default: [])
For example:
excludes all files and folder within the Sub-01 folder. This means the following directories would not be uploaded:
--force-scan Try to parse all files as DICOM files regardless of the DICM prefix. (default: False)
(might want to use --include "\*")
-g ID, --group ID The id of the group if not in the folder structure
--group-override ID Force using this group id
--ignore-unknown-tags Ignore unknown dicom tags when parsing dicom files (default: False)
By default ingest can fail if the tag is not private and has implicit VR and not in this list. These tags can be ignore with this cli option.
Flywheel uses the pydicom dictionary to determine known tags
--include PATTERN Patterns of filenames to include (default: [])
For example:
Include a single file:
Include all files of a filetype:
--include-dirs PATTERN Patterns of directories to include (default: [])
The regex wildcard means that this would include the directories like:
* OHM/101-1011
Note: when S3 bucket is configured as source, this flag does not support regex wildcard match, only “starts with.”
--load-subject PATH Load subjects from the specified file
--no-audit-log Skip uploading audit log to the target projects
--no-sessions No session level (create a session for every subject)
(default: False)
Use this flag if you do not have data to create sessions in Flywheel, but you do have subject data. Flywheel uses the corresponding subject label to create the session label.
--no-subjects No subject level (create a subject for every session)
(default: False)
Use this flag if you do not have data to create subjects in Flywheel, but you do have session data. Flywheel uses the corresponding session label to create the subject label.
--pack-acquisitions TYPE Acquisition folders only contain acquisitions of TYPE and are zipped prior to upload
- p LABEL, --project LABEL The label of the project, if not in folder structure
--project-override LABEL Force using this project label
--require-project Proceed with the ingest process only if the resolved group and project exists (default: False)
By default, Flywheel creates a new group or project if an existing group ID or project label does not match what you designated in your command or template. This means if you mistype the group id or project label, your data will be uploaded in the wrong location. Use this flag to make sure your data is only uploaded to existing groups and projects.
--root-dirs ROOT_DIRS The number of directories to discard before matching (default: 0)
--skip-existing Looks at filenames for files already in your project, and skip import of files with the same filename. (default: False)
--symlinks Follow symbolic links that resolve to directories
(default: False)


Optional argument Description
-C PATH, --config-file Specify configuration options via config file. Learn more about how to create this file.
--no-config Do NOT load the default configuration file
-y, --yes Assume the answer is yes to all prompts
--ca-certs CA_CERTS Set the effective local timezone for imports
--timezone TIMEZONE Set the effective local timezone for imports
-q, --quiet Squelch log messages to the console
-d, --debug Turn on debug logging
-v, --verbose Get more detailed output


These config options are only available when using cluster mode with the --follow argument or when using local worker.

Optional argument Description
--save-audit-logs PATH Save audit log to the specified path on the current machine
--save-deid-logs PATH Save deid log to the specified path on the current machine
--save-subjects PATH Save subjects to the specified file


These config options apply when using a cluster to ingest data.

Optional argument Description
--cluster CLUSTER Ingest cluster url (default: None)
-f, --follow Follow the progress of the ingest (default: False)


These config options are only available when using local worker (--cluster is not defined)

Optional argument Description
--jobs JOBS The number of concurrent jobs to run (e.g. scan jobs), ignored when using cluster (default: 4)
--sleep-time SECONDS Number of seconds to wait before trying to get a task (default: 1)
--max-tempfile MAX_TEMPFILE The max in-memory tempfile size, in MB, or 0 to always use disk (default: 50)The max in-memory tempfile size, in MB, or 0 to always use disk (default: 50)