README

Code utilized for the UNET model implemented in the following publication:

Automated Segmentation of Kidney Cortex and Medulla in CT Images: a Multisite Evaluation Study

Reference the above paper when parts of the repo are used.

This repo contains the model as well as the inference code to apply the model in new datasets. An example case has been added. Due to limitation in sharing the training data an example test case (case_00000_image_reo_crop.nii.gz) originating from a publicly available dataset is used.

Please note that the algorithm was developed utilizing a corticomedullary CT phase. We are not aware of publicly available datasets containing appropriate images so we opt to use an image from a contrast enhanced CT scan. Additional the test case contains abnormalities. The data used to train the algorithm originate from healthy individuals.

Running the inference code

Execute the following command:

python inference_tf.py --input_image data/case_00000_image_reo_crop.nii.gz --output_filename data/case_00000_image_reo_mask.nii.gz --model weights/weights.h5

Note that depending on how your environment is configured you might have to replace python with python3

Dependencies

Please utilize python 3.8.6 or above

See the requirements.txt for dependencies.

pip install -r requirements.txt

Issues

Report any issues utilizing the issue reporting mechanism.

Contirbutors

Timothy L Kline (Kline.Timothy@mayo.edu)

Panagiotis Korfiatis (Korfiatis.Panagiotis@mayo.edu)

Let us know if you are interested in collaboration relevant to this repository.