docs/README.md

Multi-organ segmentation in abdominal CT

Authors

Chen Shen¹, Holger R. Roth², Kazunari Misawa³, Kensaku Mori¹

1. Nagoya University, Japan

2. NVIDIA Corporation, USA

3. Aichi Cancer Center, Japan

Tags

Segmentation, Multi-organ, Abdominal

Model Description

This model uses the DiNTS model architecture searched on Medical Segmentation Decathlon Pancreas [1] and re-trained for multi-organ segmentation from abdominal CT images [2,3].

Data

This model was trained on an abdominal CT dataset in portal venous phase collected from Aichi Cancer Center in Japan. Since this is a private dataset, similar models can be trained using other public multi-organ datasets like BTCV.

For this bundle, we split the 420 cases into training, validation and testing with 300, 60 and 60 cases, respectively.

Output

8 channels

• 0: Background
• 1: Artery
• 2: Portal vein
• 3: Liver
• 4: Spleen
• 5: Stomach
• 6: Gallbladder
• 7: Pancreas

Here is an example of output.

Scores

This model achieves the following Dice score on the validation data (our own split from the whole dataset):

Mean Dice = 88.6%

MONAI Bundle Commands

In addition to the Pythonic APIs, a few command line interfaces (CLI) are provided to interact with the bundle. The CLI supports flexible use cases, such as overriding configs at runtime and predefining arguments in a file.

For more details usage instructions, visit the MONAI Bundle Configuration Page.

Execute model searching:

python -m scripts.search run --config_file configs/search.yaml

Execute multi-GPU model searching (recommended):

torchrun --nnodes=1 --nproc_per_node=8 -m scripts.search run --config_file configs/search.yaml

Execute training:

python -m monai.bundle run --config_file configs/train.yaml

Please note that if the default dataset path is not modified with the actual path in the bundle config files, you can also override it by using --dataset_dir:

python -m monai.bundle run --config_file configs/train.yaml

Override the train config to execute multi-GPU training:

torchrun --nnodes=1 --nproc_per_node=8 \
     -m scripts.search run \
     --config_file configs/search.yaml

Override the train config to execute evaluation with the trained model:

python -m monai.bundle run \
    --config_file "['configs/train.yaml','configs/evaluate.yaml']"

Execute inference:

python -m monai.bundle run --config_file configs/inference.yaml

Export checkpoint for TorchScript:

python -m monai.bundle ckpt_export network_def --filepath models/model.ts --ckpt_file models/model.pt --meta_file configs/metadata.json --config_file configs/inference.yaml

Execute inference with the TensorRT model:

python -m monai.bundle run --config_file "['configs/inference.yaml', 'configs/inference_trt.yaml']"

References

[1] He, Y., Yang, D., Roth, H., Zhao, C. and Xu, D., 2021. Dints: Differentiable neural network topology search for 3d medical image segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 5841-5850).

​ [2] Roth, Holger R., et al. "A multi-scale pyramid of 3D fully convolutional networks for abdominal multi-organ segmentation." International conference on medical image computing and computer-assisted intervention. Springer, Cham, 2018. ​

[3] Shen, Chen, et al. "Effective hyperparameter optimization with proxy data for multi-organ segmentation." Medical Imaging 2022: Image Processing. Vol. 12032. SPIE, 2022.

License

The Licensee is not allowed to distribute or make the model to any third party, either for free or for a fee. Reverse engineering of the model is not allowed. This includes, but is not limited to, providing the model as part of a commercial offering, sharing the model on a public or private network, or making the model available for download on the Internet.