fw-gear-nnunetv1 0.1.2

Run inference by the nnUnet v1 model.

nnunetv1

Note: Unit tests have not been added to this repository yet.

Summary

Run inference by the nnUnet model [1, 2] on an input NIfTI file.

Testing datasets, from http://medicaldecathlon.com/ [3]:

• Task04_Hippocampus.tar (a 3-D dateset)
• Task05_Prostate.tar (a 4-D dataset)

Cite

[1] Isensee, F., Jaeger, P.F., Kohl, S.A.A. et al. (2020). nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat Methods. https://doi.org/10.1038/s41592-020-01008-z.

[2] Isensee, F., Jäger, P. F., Kohl, S. A. A., Petersen, J., & Maier-Hein, K. H. (2021). pretrained models for 3D semantic image segmentation with nnU-Net (2.1). Zenodo. https://doi.org/10.5281/zenodo.4485926.

[3] Simpson, A. L., Antonelli, M., Bakas, S., Bilello, M., Farahani, K., Van Ginneken, B., ... & Cardoso, M. J. (2019). A large annotated medical image dataset for the development and evaluation of segmentation algorithms. arXiv preprint arXiv:1902.09063.

License

The nnUnet source code is under an Apache License 2.0; the remainder of this gear is under MIT License.

Classification

Category: Analysis

Gear Level:

• Project
• Subject
• Session
• Acquisition
• Analysis

---

[[TOC]]

---

Inputs

• modality_0

  • Name: modality_0
  • Type: .nii.gz
  • Optional: false
  • Classification: file
  • Description: 3D or 4D NIfTI file.
  • Notes: If 4D, should be only modality input.

• modality_1

  • Name: modality_1
  • Type: .nii.gz
  • Optional: true
  • Classification: file
  • Description: 3D NIfTI file.
  • Notes: If this exists, all other inputs should also be 3D.

• modality_2

  • Name: modality_2
  • Type: .nii.gz
  • Optional: true
  • Classification: file
  • Description: 3D NIfTI file.
  • Notes: If this exists, all other inputs should also be 3D.

• modality_3

  • Name: modality_3
  • Type: .nii.gz
  • Optional: true
  • Classification: file
  • Description: 3D NIfTI file.
  • Notes: If this exists, all other inputs should also be 3D.

Config

• pretrained_model

  • type: str
  • Description: Pre-trained model to use for inference. 10 options given for Task models 00-10 [2].

• debug

  • type: bool
  • Description: Whether to include debug statements in the job logs.

Output Files

• prediction_time.txt
• postprocessing.json
• plans.pkl
• {input NIfTI file base name}__{model name}.nii.gz

Workflow

1. Upload file(s) to container.
2. Select file(s) as input(s) to gear.
3. Specify which model to apply inference with, as well as any other config selections and run.
4. Gear runs inference with selected model on inputs and places results into into new Analysis container.

Contributing

[For more information about how to get started contributing to that gear, checkout CONTRIBUTING.md.]