NYU Center of Data Science Capstone Project (Fall 2021) by Claire Ellison-Chen, Kallen Xu, and Yupei Zhou.
loadDataset.py loads the NYU Langone dataset on Bigpurple and splits it into training/validation/test dataset.
preprocess.py applies necessary preprocessing steps including cropping/padding, resampling and z-normalization to the data.
train.py and test.py contains scripts to train and test the 3D Unet model.
lossOption.py contains several loss functions for model training.
metricOption.py contains several metrics for model evaluation.
visualization_result.py includes python scripts for visualization.
util.py contains other relevant python scripts.
example_for_script.ipynb is an end-to-end example of loading and preprocessing data, training the model and testing the model.
patient_eval.ipynb creates visualization using trained models on patient data.