Deep Learning for Automatic Segmentation of Vestibular Schwannoma: A Retrospective Study from Multi-Centre Routine MRI -- Deep learning models
Description
These are zipped folders containing all models trained with nnU-Net for the journal publication:
Deep Learning for Automatic Segmentation of Vestibular Schwannoma: A Retrospective Study from Multi-Centre Routine MRI
The Zenodo upload contains the following files:
- Multi-Centre-Routine-Clinical-(MC-RC)-models.zip
- Models trained on the MC-RC dataset
- Single-Centre-Gamma-Knife-(SC-GK)-models.zip
- Models trained on the SC-GK dataset
- MC-RC+SC-GK-models.zip
- Models trained on both datasets
- example_input_images.zip
- example images to test the inference
To run inference from a Linux command line, follow these steps:
1. install the nnU-Net (v2) python package. This can be done with the following command:
pip install nnunetv2
2. unzip the model folders
3. set the environment variable `nUNet_results` to the path that contains the unzipped model folders (e.g. Dataset910_VSMCRCT1, Dataset911_VSMCRCT2, etc.). For example you can use the following command:
export nnUNet_results="/home/username/Multi-Centre-Routine-Clinical-(MC-RC)-models/"
4. follow the model-specific instructions under <model-folder>/inference_instructions.txt
Make sure to replace INPUT_FOLDER, OUTPUT_FOLDER, etc. in the commands with valid paths.
The final post-processing command starting with nnUNetv2_apply_postprocessing should be omitted.