FLAMeS Inference

This package reduces the complexity of using FLAMeS, a model for multiple sclerosis lesion segmentation from a single FLAIR image, into one intuitive command.

FLAMeS official website.

Installation

git clone https://github.com/Gad-MA/flames-inference.git
cd flames-inference
pip install -e .

Usage

Place all the images you want to segment in a folder.

Make sure that all images follow the following naming format: CASE_IDENTIFIER_0000.nii.gz

Example input image names:

• brain_001_0000.nii.gz
• brain_002_0000.nii.gz

Then run the following commands:

export MPLBACKEND=Agg # This command sets an environment variable in Bash. If you are using a different shell, use its equivalent command instead.
flames-inference --input <input_folder_with_nifti_images> --output <output_folder>

Using Google Colab

If you use colab, check the "FLAMeS_Inference_in_Colab.ipynb" notebook for detailed instructions.

Help

Run the following command to view all available options and capabilities:

flames-inference --help