Manual correction

This repository contains scripts for the manual correction of spinal cord labels.

Currently supported labels are:

• spinal cord segmentation
• gray matter segmentation
• lesion segmentation (e.g., MS or SCI lesions)
• disc labels
• compression labels
• ponto-medullary junction (PMJ) label
• rootlets segmentation
• centerline

[!NOTE] We greatly appreciate feedback and suggestions for improvement. Feel free to open an issue and report bugs, suggest new features or ask questions.

Table of content

• 1. Dependencies
• 2. Installation
• 3. Usage
  • manual_correction.py
  • package_for_correction.py
  • copy_files_to_derivatives.py
• 4. Examples

1. Dependencies

• Spinal Cord Toolbox (SCT)
• FSLeyes or ITK-SNAP
• Python 3.8

2. Installation

Download the repository:

git clone https://github.com/spinalcordtoolbox/manual-correction.git
cd manual-correction

Create a virtual environment and install the required packages:

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Alternatively, you can use existing SCT's conda environment:

source ${SCT_DIR}/python/etc/profile.d/conda.sh
conda activate venv_sct

3. Usage

[!Important] All scripts currently assume BIDS-compliant data. For more information about the BIDS standard, please visit http://bids.neuroimaging.io.

manual_correction.py

This script loops across subjects listed in the YAML file and opens a viewer to correct the labels (segmentations, disc labels, etc.).

python manual_correction.py -path-img <INPUT_PATH> -config <CONFIG_FILE>

• INPUT_PATH: Path to the BIDS-compliant folder with data to be corrected.
• CONFIG_FILE: YAML file that lists images to be corrected. This YAML file can be generated by SCT's QC HTML report or created manually.

For full help, please run: python manual_correction.py -h.

package_for_correction.py

If the manual correction is done by someone else than the person doing the processing, this script packages the processed images and existing labels (segmentation, disc labels, etc.) and creates a single ZIP file that can conveniently be sent to collaborators.

python package_for_correction.py -path-in <INPUT_PATH> -config <CONFIG_FILE>

• INPUT_PATH: Path to the BIDS-compliant folder with data to be packaged.
• CONFIG_FILE: YAML file that lists images to be packaged. This YAML file can be generated by SCT's QC HTML report or created manually.

For full help, please run: python package_for_correction.py -h.

copy_files_to_derivatives.py

This script copies manually corrected labels (segmentations, disc labels, etc.) from your local derivatives/labels folder to the already existing dataset's derivatives/labels folder.

python copy_files_to_derivatives.py -path-in <INPUT_PATH> -path-out <OUTPUT_PATH>

• INPUT_PATH: Path to the BIDS-compliant folder with manually corrected labels.
• OUTPUT_PATH: Path to the BIDS-compliant folder where manually corrected files will be copied.

For full help, please run: python copy_files_to_derivatives.py -h.

4. Examples

See wiki