yu-to-msk / CT-Intensity-Calibration-Phantom-Segmentation

CT-Intensity-Calibration-Phantom-Segmentation

This software segments the intensity calibration phantom (B-MAS200, Kyoto Kagaku, Kyoto, Japan) from CT images. The segmentation model was built from convolutional neural network using a Bayesian U-net architecture (https://github.com/yuta-hi/bayesian_unet) and was trained on manual segmentations of 40 cases.

Requirements

• Python 3
• keras 2.2.3
• CPU or NVIDIA GPU + CUDA CuDNN

For details, please see requirements.txt

Reference

Preprint of the paper using this model can be found at (http://arxiv.org/abs/2012.11151).
Journal full paper can be found at (https://doi.org/10.1007/s11548-021-02345-w)

When using this model, please cite

• Uemura, K., Otake, Y., Takao, M. et al. Automated segmentation of an intensity calibration phantom in clinical CT images using a convolutional neural network. Int J CARS (2021). https://doi.org/10.1007/s11548-021-02345-w

Usage

Inference Code for Phantom Segmetation using Bayesian U-net.

Inputs:

• '-i' or '--in_dir' : Directory including MHD images (default: data)
• '-o' or '--out_dir' : Directory to save segmentation results (default: results)
• '-u', '--uncert_ok' : Flag to save the uncertainty or not (default: False)

Examples:

• python phantom_segmentation.py -i "./data" -o "./labels" -u
• python phantom_segmentation.py --in_dir "./data" --out_dir "./labels" --uncert_ok
• python phantom_segmentation.py --in_dir "./data" --out_dir "./labels"

Licence

This software can be used for research purpose or for educational purpose. For commercial use, please contact the Imaging-based Computational Biomedicine Lab, Nara Institute for Science and Technology, Japan.

For details, please see LICENCE.txt