Segment-Flow

Nextflow pipeline for deep learning segmentation models

Usage

To use this pipeline, you need to have Nextflow availabile. To install Nextflow, see their installation guide.

To run the pipeline locally, the profiles/local.conf profile should be all that is needed. To run the pipeline on a cluster, you will need to create a profile for your cluster. See the Nextflow documentation and the AIoD wiki for more information.

Parameters

To get the arguments for the pipeline with a description for each, run the following command:

nextflow run FrancisCrickInstitute/Segment-Flow --help

The following sections will provide some extra information about key parameters.

Input CSV (--img_dir)

Typically, this CSV is generated by the Napari plugin. To use this pipeline standalone, you can create a CSV with the following columns:

• img_path: Path to the image file
• num_slices: Number of slices in the image
• height: Height of the image (in pixels)
• width: Width of the image (in pixels)
• channels: Number of channels in the image

Additional columns are created by the pipeline and are not required in the input CSV.

Model Name & Version

TODO AFTER PIPELINE MODEL HANDLING REVISION